Startups. I can't help myself. Trouble ahead for SaaS Sun, 03 Dec 2017 00:00:00 +0000 From a user demand angle, the SaaS business model is built on a simplification advantage: A SaaS tool gives you the functionality you need to do your job, ready to use and hassle-free. You don’t have to install anything, or make sure it’s updated, or manage servers, tend to security, make sure it scales, etc. In return, they charge a monthly fee that is tied to the the ongoing value-added to the user. The per person logic is a favorite, reflecting this view that they’re optimizing each users workflow continuously, but other feature-based pricing models exist.

From a supply angle, it seems like the proliferation of SaaS products has happened because of how easy and inexpensive it has become to create such a tool, even for rather narrow verticals:

  • Open Source. Abundance of Open Source software, libraries and API integration code.
  • Inexpensive infrastructure. Lots of infrastructure as a service (IaaS) and platforms as a service (PaaS) offered by Google, Amazon, Heroku, etc.
  • Automated development operations. Releasing, configuration (infrastructure as code) and monitoring are increasingly done via automation.

However, I believe that these forces will also ultimately be the demise of SaaS as the king of B2B software models. In the future, I believe, many users will choose the hosting of their choice and auto-deploy software from Open-Source or commercial repositories. Infrastructure operators (IaaS and PaaS) will probably bundle this setup of automatic deploying and updating etc, like Heroku autodeploy. The easier they make it, the more you’ll spend. In a way, it will be the natural evolution of the one-click install that many cloud providers offer for popular software like Magento (eCommerce) or Drupal (MS), just in a model that is ready to scale and easy to manage and update.

The big wins will be more transparent pricing, lower prices and more control.

SaaS may fight back and replicate pricing models tuned to infrastructure usage. They can and will keep investing in differentiating functionality that is hard to replicate under a distributed model, especially for niches. Big data, machine learning models will also bring advantages to the players controlling most of the data when these times arrive. Complex models may survive in SaaS, but the future clearly will bring a lot of hardship to this model.

Open Source + automated dev operations + IaaS > SaaS.

Consumer hardware Fri, 29 Sep 2017 00:00:00 +0000
  • Want: An indestructible and nice looking electric bicycle (Sep 2017). Found Budnitz model E. (Sep 2017).
  • Want: A tabletop dishwasher for kitchenware (plates, cutlery) that works just like bar glass washers. Loading a dishwasher until its full, waiting for the cleaning and unloading it again frequently seems less efficient than cleaning the wares one by one, if there's a machine that makes the washing fast enough (Nov 2017).
  • Want: a gamified toothbrush, that gives me incentives to brush my teeth more often and better (Dec 2016).
  • 2025: Finally, someone produces a very light, internet-connected device for voice and text that has a battery lasting a week (Dec 2016)
  • ]]>
    Unmanned Vehicles Sun, 26 Feb 2017 00:00:00 +0000 In the middle of 2016, me and my partner Torben decided to study a few business spaces. The criteria was to pick the idea we loved the most out of those where our competences and resources combined could result in a differentiated offering.

    We both loved the potential to tackle the software stack fragmentation in the unmanned vehicles market. Unmanned vehicles in this context includes air, ground, surface and the underwater kind. We developed a strategic document with two main references for strategic positioning. I have included them here for your reference, click the links to download!

    Market segmentation

    In the market segmentation analysis, we explored two axis. A vertical axis for activities, in terms of the type interaction with the environment required by the application. We started with Tinkering (none required), then Playing (in an environment), then Acquiring data (sensing the environment), then Surveillance (acting on repeatedly sensed data), then Transportation (of things in the environment) and finally Actuation (changing the environment). The other axis has the medium of operation, from Air (UAV), to Ground (UGV), to Surface (USV), to Underwater (UUV). It's also split, where we found appropriate, into Public (eg. streets or a lake) or Private (eg. your house or a warehouse). For each of the opportunities created in the intersection of axis we then added examples, and assessed the potential according to a number of criteria.

    Software Stack analysis

    We also produced a preliminary analysis of different sw stacks for operating unmanned vehicles. More important than the data, which is becoming rapidly stale, is the taxonomy. It starts with 3 top-level pieces of the stack. The first is an Onboard Command & Control software, which the sw that is inside each drone or each vehicle. The second is Communication, which includes all about the protocol and application-level command support of the stack. The last one is the Remote Command & Control, usually known as the Ground station. It's the tool a human or a team use to issue missions, plans, actions to the vehicles, to observe the behaviour of the vehicles, and to handle payload.

    FYI, we ended up pursuing something else. But this space is truly great. I have to thank the lab LSTS all the inspiration and support they gave me in this data compilation and analysis. Ping me for any question you may have!

    Software Sat, 31 Dec 2016 20:03:36 +0000
  • Want: iOS print as PDF for the share functionality (Sep 2017). Released in iOS 11 (Nov 2017).
  • Want: A setting that would make my browser accept and persist cookies only for sites for which I have a password. That is, I don’t mind that a website stores cookies to make my life easier, but I don’t want to help others track me (Sep 2017)
  • 2020: There is a way for non-techies to program connected applications. More than 1 million users create their custom automated workflows apps in a platform (Dec 2016)
  • ]]>
    Robots Sat, 31 Dec 2016 20:02:05 +0000
  • 2020: Rovers for last mile deliveries are rolled out in the whole of one city for all customers by at least one big brand. The city has more than 1 million inhabitants covered by this service. The service can be something like pizza hut, 7 eleven or amazon (Dec 2016)
  • 2020: Air delivery drones are still not a thing for last mile delivery in the conditions defined for rovers above (Dec 2016)
  • 2030: General purpose AI is still not a thing. Machines can’t take care of themselves autonomously. Machines and algorithms can still only learn what we tell them to learn. Even when they can perform generalization, they can only do it in types of problems we tell them to generalize (Dec 2016)
  • ]]>
    Communications Sat, 31 Dec 2016 19:57:52 +0000
  • 2025: Software SIMs are commonplace in smartphones, or whatever replaces them. The pressure of millions of connected devices, MVNOs and other hardware makers and software platforms resulted in that situation (Dec 2016)
  • 2030: There is a working mesh network that is governed by an open protocol spanning millions of nodes that are built into mobile devices like phones or watches. Coverage doesn’t reach the whole US, but some personal devices in cities effectively run critical services (like personal communication) on them (Dec 2016)
  • ]]>
    Human Interfaces Sat, 31 Dec 2016 19:55:31 +0000
  • 2017: VR and AR are not a thing. No single player sells 10m units to consumers (excludes industry) (Dec 2016)
  • 2020: VR and AR are still not a thing. No single player sells 10m units to consumers (excludes industry) (Dec 2016)
  • 2025: Machines can understand 95%+ speakers of a language better than humans with little or no custom training for that user, even in street talk (Dec 2016)
  • 2030: Computer-brain interface implant debates have been on the front page of the major media outlets. There’s at least a thousand people using them outside pure medical applications (Dec 2016)
  • ]]>
    Government Sat, 31 Dec 2016 19:52:08 +0000
  • 2030: AB Testing of governments is now possible. At least one country has 2 duplicated and parallel governments that compete with each other for at least one service. An Example is 2 competing tax authorities, polices, health systems, school systems. Citizens can subscribe to either, with duties rights according to their choice. Likely there is some lock-in when choosing. (Dec 2016) Update: It is required that the services are 100% public, that is, they are ran by the government and not by private companies regulated by the state. It is also required that the funding of the services, that include public workers, depends on the peoples choice. (Dec 2017)
  • 2030: Democracy is strengthened in several countries that now have a 4 "independent" power: Executive, Legislative, Judiciary and Informational. The Information body includes the statistics bureaus, labelling norms and data verification. It overall manages consent in society, including actions of the 3 remaining powers (Jan 2017)
  • 2035: In the years between Brexit announcement (year 2016) and 2035, per-capita GDP (PPP) in the UK will have grown as much or more than EU average (exc. UK) and of DE, FR, IT. (Jan2018).
  • Want: A Transaction Tax that replaces all taxes. This TT is collected by the bank whenever a transaction takes place. Its made of 2 sub-taxes collected at once TT=ST+RT. ST is the Sender Tax and is determined by the Country of the sender of the money; RT is the Receiver Tax and is determined by the Country of where the receiver provides the service; In case the Service isn't rendered in the same country where the receiver actually is, then the Receiver Tax can be compounded by a factor that reflects this. If the Sender is a company and the Receiver a person, then TT is the income tax. If the Sender and Receiver are the same person, we're just moving money around (no tax). But if Sender and Receiver is the same corporation, then TT is a tax for an intra-group transaction. If Sender is a person and Receiver a company, TT is a consumer Sale Tax. If Sender and Receiver are companies, TT is a B2B Sale Tax. Etc. To avoid money flowing to the parallel economy, TT for cash withdrawals is big. TT and ST and RT can vary per activity of Sender and Receiver and by the type of product or service rendered.
  • ]]>
    Business models Fri, 30 Dec 2016 00:00:00 +0000
  • 2017: Apple continues to hold >90% of the profit of the smartphone industry (Dec 2016). Happened.
  • 2017: Apple becomes the number 1 watch maker in the world by revenue (Dec 2016). Tim Cook announced it on 2017-09-12 (September 2017)
  • 2018: Apple crosses the 1 trillion market capitalisation (Nov 2017)
  • 2020: ICOs didn't kill the VC business. VCs still exist and they're still how startups get funded. They still deploy the same kind of money they were deploying in 2016, or more. If ICOs and smart contracts have equity-like legal power then VCs invest through them. There are a few ICOs per ecosystem (storage, computing, services), #ICOs << #Startups using ICOs. (Nov 2017)
  • 2020: There is at least one worker digital union (syndicate) valued at 1b. These digital unions startups are online aggregators that tilt the bargaining power of individuals vs. big platforms. (Dec 2016). Update: in big platforms I include non-digital companies, it can be a big company. (Set 2017)
  • 2020: There is at least one consumer digital union (cooperative) valued at 1b. These digital unions startups are online aggregators that tilt the bargaining power of individuals vs. big platforms. (Dec 2016)
  • 2025: Verifiable video and photos. A service exist that verifies when and where a video or photo was taken. Such service is used by a top newspaper of the world (WaPost, Times, NYT, etc). It’s composed by an app which records video or pictures, and that guarantees that the media comes from the camera hardware and that location is from the GPS chip, and not some SW hack. The app then calls a service that stores the media and meta data or a hash of the media and metadata. The service can be a blockchain based service. Separate apps and services allow verifying any video posted anywhere on the internet. (Dec 2017)
  • 2025: CV on the blockchain. A decentralised social network for CVs has been launched and attracted more than 10m users. Users are incentivised to join by awarding them tokens of the network. They also get more tokens if they remain on the network, potentially tiered by network size, age, or network value. Miners also get tokens. Tokens are used to access the CV data, via API or via the app. The more data of a person or persons you need to access, the more you pay. Users can block certain info, and maybe get less tokens awards. (Feb 2018)
  • 2025: Bitcoin (BTC) reaches USD 50.000, Ethereum (ETH) reaches USD 5.000. They don't need to be at those values by 2025, just reach it before 2025. (Jan 2018)
  • 2025: Finally, your digital ID now stores passwords for most things, including websites, apps, services, banks, etc. You need only a handful (5), at most. Maybe it depends on a hardware device (Dec 2016)
  • 2025: Digital debt is a thing. There is at least one pure-digital debt issuer and one online-first credit worthiness scoring agency. Online identities track your score throughout the internet. They are available for contracting to the majority of clients of a major player with eCommerce or services operations. Can be Amazon, Apple, Google, AliBaba (Dec 2016)
  • 2025: Region lock for content fades way. Platforms like Netflix, Spotify and iTunes have grown, and international trade deals have helped trigger catalogues that are 90% equal for access in different regions. Has to include at least US and EU (Dec 2016)
  • 2025: Open-work on Open-source. At least one company worth >$1b opens their code and pays developers across the world by reputation of their commits within their open source software trunk, which can cover all or part of the company. Incentives exist for great contributors, paying them an on-going allowance for some months after their relevant contribution. At least 20% of dev budget is paid out this way. (Oct 2017)
  • 2025: Uber has either stopped developing its autonomous hardware for small cars in-house, or became a car maker (Dec 2016)
  • 2030: Amazon, Facebook or Google's market value shows a downwards trend until 2030. The fact that their market value is very dependent on a dominant position will attract growing government oversight. Amazon is complex and produces little profit, which makes investment dependent on the market believing on continued growth and market domination - hard to maintain conditions. Google and Facebook make the majority of money from a single service (advertising), and the online ads industry can be changed overnight by governments action or tech disruption. (Dec 2017)
  • 2030: In several verticals, SaaS no longer dominates. Ex: accounting, corporate communication, etc. Subscriptions lost to Software that is automatically deployed from a open-source/ private repository to the users infrastructure, which is being managed in a consumption/ subscription model. Recipes automate updating and security, scaling etc (the devops). (Nov 2017)
  • 2030: Co-living, co-owning and house time-sharing is big. People are mobile and have contracts that guarantee shelter. The population in the West excluding immigration is still declining fast relative to the amount of housing available, and one these flexible rental services have more than a million active contracts (Dec 2016)
  • 2030: Car sharing and car hailing services provide 50% of more of all intercity trips in one of the largest 100 cities in the world (Dec 2016)
  • ]]>
    Design Sprint Tue, 29 Nov 2016 00:00:00 +0000 A sprint is a 1 week super-focused exercise where a team is gathered to work continuously on a problem. Sprints should be used to test if a concept or product will work prior to making a big resource commitment in people or money. Cristina Fonseca, founder of TalkDesk, recommend me to look at this concept. I did it on my current company and really liked it.

    Jake Knapp developed a sprint process for Google Ventures that he then executed many times with well known startups. Jake also led sprints at Google for projects inside Gmail and Google X.

    From Monday to Friday, each team member participating in the sprint should be 100% focused on the problem presented and avoid all distractions. No phones or laptops except where needed, and in short breaks. There is a maximum of 7 team members. You need a Decision Maker that takes part in the process as an individual, and in the end has the ultimate power to decide outputs. Another of those seven is the Facilitator who is in charge of schedules and the flow. You need to bring relevant people from each area of the company. Due to the level of focus needed, 6 hours should be the limit per day.

    The broad tasks and schedule are defined as such by Knapp on his book but you can also check it on wikipedia.

    Marketing campaigns Thu, 28 Apr 2016 00:00:00 +0000 If you want to do campaigns in marketing, you have a lot to choose from. Online, offline, new customers, repeat customers, there’s thousands of different campaign types. You will only know what’s best after you’ve tested them, optimised them. So there will be a lot of trial and error involved. And even after you nail it, things may change in time, so you’ll want to revise results.

    However, you should have a strong initial hypothesis about what should work and what shouldn’t. This will make your search more targeted and easy. I’ve made a small survey of the different types of marketing campaigns, and some of the qualities of each, to help you with that.

    You can find the whole survey, including the classification here. If you want to play with it and customise the attributes for your industry, you should download the file as an Excel doc, or duplicate the sheet into your drive.

    Below I explain the file, which I simplified for this example.

    Channels and their campaigns

    1. CRM. All campaigns targeting customers who have purchased at least once in all platform. So all converts coming from here are for repurchases. As a corollary, all other remainder channels and campaign types are for Customer Acquisition and converts are new purchases.
    2. Online Marketing. All performance customer acquisition campaigns done online. That is, scalable online channels with a pure intent to convert new customers.
    3. Offline B2B. All offline campaigns targeted at acquiring new customers in a business setting.
    4. Offline B2C. All offline campaigns targeted at acquiring new customers in a consumer setting.
    5. PR. All campaigns, online or offline, whose intent is to educate and engage new customers through a channel with a big authority (magazines, famous people), most likely without incentives or discounts.
    6. CONTENT. All online campaigns that focus on the information itself, and not the medium or person, as the means to convert new customers.


    1. Demo: the exposure of the target to the product
    2. Target relevance: who are we reaching
    3. Push / Pull: If we are being pushy, of if the customer was already looking for us
    4. Management: How easy is this?
    5. Direct / Indirect: Are we going to the customer as ourselves, or through a partner?
    6. Viral potential: is this sharable?
    7. Reach: How many people will this reach?
    8. CVR: How much will this convert, per person?
    9. CAC: What’s the cost of each conversion?
    10. Your competitor: Are you competitors doing these things? (An indication, not necessarily a confirmation of this being good)
    11. Your segment (several): How do you expect each of your segments to respond to this?
    12. Average: The attributes above are not perfectly MECE, especially Reach, CVR, CAC vs. all the others. This is because we want a nice overview of the campaigns, not a perfect score. The score is really just a fetish.
    13. Priority: This is your priority. What you want to do now, what you don’t want to do, and what you may want to try.


    • Output KPIs. These are the lagging indicators which I frequently use for all my campaigns everywhere. They are MECE and well studied: number of Converts, Customer Acquisition Cost (CAC) and Customer Lifetime Value (LTV).
    • Performance KPIs. These are the leading indicators which can be used to control larger teams and how each campaign is doing to contribute towards the output KPIs.

    Remember this is a particular survey of channels and campaigns. You should always re-interpret and discuss within your team what is the impact of each campaign in all of those attributes I picked, or others you find more relevant.

    Pricing Sat, 23 Apr 2016 22:45:28 +0000 You built a product. Now its time to make some money. You want a healthy margin, but you don’t want to over charge and ruin the business long term. How to price properly?

    One way to do it is to build your pricing thermometer. The price thermometer is the following:

    || True Value
    || Perceived Value
    ||       |
    || buyer incentive
    ||       |
    || Price
    ||       |
    || seller incentive
    ||       |
    || Cost


    • True value is the value that your product generates.
      • If you are selling an accounting software that allows a company to save $1.000 in tax, then your true value is $1.000. This value is, frequently, hard to determine. Many segments of the market have different uses for a product, which unlocks various different True Values for each of them.
    • Perceived value is the value that the customer understands.
      • It takes into account the marketing you do, but also the competitors and replacements available. Let’s say that your accounting software has 3 packages: HR, Procurement and Sales. Let’s say the software will save a customer $500 in HR, $490 in Procurement, and only $10 in Sales taxes. Because the software is quite complex, you don’t advertise the Sales taxes savings prominently, and you don’t know if a regular customer will be able to use more than 50% of the Procurement features. Therefore the perceived value is actually only $500+50%*$490 = $745. Finally, if there is a direct competitor selling a similar software for $695, then obviously this will be the perceived value of your product as well. If their software is not exactly similar, you should compare each individual feature and its value.
    • Buyer Incentive = Perceived value - Price
    • Price. The price you are selling the product.
    • Seller Incentive = Price - Cost
    • Cost of Product. This is the cost of goods or services sold.

    Basic stuff

    • If Price = Cost, you make no money at all. If Price = Perceived Value, the customer is getting no benefit: he is paying as much for the product as he’ll benefit from it.
    • You should place price in between cost and the perceived value. The higher the price, the more margin you will make. Your incentive to sell equals Price minus Cost. The lower the price, the more benefit a user will extract. The customers incentive to buy equals Perceived Value minus Price.
    • Generically, the higher the margin between perceived value and cost, the most likely an industry is likely to develop. Because Seller Incentive + Buyer Incentive = Perceived Value - Cost.

    More. This notion of pricing using the thermometer is quite static. It’s very important when you want to build a strong brand and product, and a long term relationship with the client. In a dynamic market, it takes time to educate your customers on how to extract the true value of your product or service. Profit is what truly matters, but if you want to maximise the revenue you get at each moment, you may need to do a bit of tinkering in Pricing.

    A quite simple and quick way to determine optimal price in a multi-customer setting is to start up and slowly go down. So, you start with a price near the maximum perceived value, say 100, and you get to sell 20 units. Then you drop to 90 and maybe you get 25 customers. You keep going down, and calculating Revenue = Price x Quantity.. At some point you reach a maximum, and that’s your optimal price to maximize revenue. You can do the same with margin of course, what you need to maximize instead is Margin = Quantity x (Price - Cost). Note that in digital software and many online services the marginal cost of serving another customer is zero, therefore maximising Revenue is similar to maximising margin.

    GOTChA Chart Sun, 17 Apr 2016 00:00:00 +0000 People use reminders, notes, calendar items and other tools to organize themselves.

    When I spent some university research time in California, I got used to organising myself using a GOTChA chart. This was proposed by my supervisor and amazing professor Richard Murray, from the Control and Dynamic Systems department at Caltech. Its main use is to track projects, though I propose you use it to keep track of your general responsibilities, in tasks big or small.

    A GOTChA chart has four main elements, as presented by Richard himself (slightly modified)]:

    • Goals. A high level description of what you want to accomplish, in plain English. Your goals should give a clear description of what you hope to accomplish in the overall timeframe of a project. A sample goal is “Increase sales and profitability of the company in the current year”.
    • Objectives. A concrete specification of what you want to accomplish, with numbers and dates. It must be simple to assess whether or not each objective has been completed. They should support the overall goals, but should be much more specific and concrete. A sample objective is “Reach an average rate of 50 new customers per day by the end of December, and for the whole of the month.”.
    • Technical Challenges. This is a list of the hard things of accomplishing your goals and objectives. The list should fairly short (~5) items and focus on those parts of the problem that are the true showstoppers. A sample technical challenge is “Currently, customers acquired through offline channels (eg. Flyers) don’t come back often enough, making each customer acquired a pain on profitability”.
    • Approach. A list of activities or strategies that you plan to implement to overcome the technical challenges. These activities should provide the justification for why you think you can achieve your goals and objectives in the face of the technical challenges you have described. An example would be “Do offline campaigns only at locations where we are fairly certain that our target customer is willing to interact with us”.

    I typically read (and present) my gotcha chart in two ways:

    • as a 2x2 table, which segments items per category. This is great to do a progress review or to present a project as a whole to a new team member.
    • as a tree, which groups items by their links to others. This is better to dive in in an analysis and to actually solve objectives one by one.

    A table would look like this:

    • Goal 1
    • Goal 2
    Technical Challenges
    • TC 1
    • TC 2
    • Objective 1
    • Objective 2
    • Approach 1
    • Approach 2

    A tree would look like this:

    • Goal 1
      • Objective 1
        • Technical Challenge 1
          • Approach 1
        • Technical Challenge 2
          • Approach 1
          • Approach 2
      • Objective 2
        • Technical Challenge 1
          • Approach 1
    • Goal 2
      • Objective 1
        • Technical Challenge 1
          • Approach 1
          • Approach 2
    M.E.C.E. Sat, 16 Apr 2016 19:08:38 +0000 I try to make all my arguments mutually exclusive and collective exhaustive. M.E.C.E in short. I learned it in consulting. What does M.E.C.E. mean? Its means that I try to avoid individual propositions to overlap between themselves and at the same time I try to address the whole, not a part of it.

    One example: When asked to provide an accurate description of a tree, one answer can be “A tree has a trunk, branches and leaves. Upon this brown and green base, you have bright and colourful flowers and fruits.”

    We can see that there are at least two problems with this description.

    • a trunk, branches and leaves are generally brown and green. Sure. But fruits can be green too. The green is not exclusive to trunk, branches and leaves. And colourful is definitely not exclusive to flowers and fruits. So the description terms as they stand are not mutually exclusive.
    • there is no mention of roots. Therefore the collective description is not exhaustive.

    The first problem is the violation of mutual exclusiveness (M.E.). It happens when we use mix different classes of attributes to describe the tree. In my example I mixed color and morphology. Why does using different classes break M.E.C.E.? Because if one set of attributes is already collectively exhaustive and covers the whole, then any other different attribute will be overlapping that whole that you just covered and therefore not be mutually exclusive with the first attribute!

    To create the example, I just described what came to my mind. If you wanted to do a proper description for your arguments, a M.E.C.E. one, you should instead start picking attributes from available observation points:

    • What makes X. In the tree example, we’re talking about the morphology: roots, trunk, branches, etc. We could also pick a more detailed level, like the types of chemical compounds that exist on a tree.
    • What does X look like. In the tree example, we can use attributes like color, but also different ones like height, or smell and texture.
    • What are the dependencies of X. In this example, it can be the sun, water, nutrients in the soil, or the wind to carry the seeds.
    • The evolution of X. For a tree, we’d be talking about the seed stage, sprout, full tree formation, or different seasons.
    • The dependents of X. In the example, it could be ashes, another tree, fertiliser, food, carbon. This one is tricky, ok. You get the picture.

    Mutual exclusiveness helps you compartmentalise. Why is it good? When you are trying to reach some conclusion with your description, you want each piece of the argument to have standalone credibility. Because even if people challenge one item, that is unlikely to affect others and your conclusion may still hold. In the example above, if you as my listener pointed that some trees have flowers and fruits that are are not at all bright and colourful, then not only would I have to amend it, but also correct the reference of the brown and green trunk and leaves being the color base of the tree relative to the flours and fruits. My whole description just fell apart. Instead if your propositions are mutually exclusive, your job is much cleaner.

    The second problem is the violation of collective exhaustiveness (C.E.). It happens when you leave stuff unmentioned. Not mentioning roots makes your description incomplete, as they may be important. And rightly so - of course they are!

    Therefore a better description would be: “A tree has roots, a trunk, branches, leaves, flowers and fruits.” This is way much more M.E.C.E. Now, we know that each additional attribute we want to mention will not be M.E.C.E. with the previous ones, but we may want to use them anyway to enrich the argument. Let’s say color. So while we can’t make color M.E.C.E. with morphology, we can at least have the color description be M.E.C.E. within each of the morphology items individually: “Roots are generally brown, and so are trunks and branches. Leaves are green, yellow, red or brown. Fruits can have those same colours. Flowers too, but also white, purple, and pretty much all colours of the rainbow.”. Now, brown is obviously not exclusive to roots, but within roots, we only mentioned brown once. Within leaves, we presented all and independent options for colours too: green, yellow, red and brown. Etc.

    For a more mathematical explanation, consider your tree to be a collection of elements. And a description is no more than naming different logic sub-sets in the finite collection of elements that represents the tree. Now, for all purposes of a logical or mathematical analysis, you make it easier if all sets are independent, and you definitely don’t want to leave any element behind. Therefore being M.E.C.E. is the equivalent of turning your Venn diagram into a nice puzzle and not missing any piece. Of course, for each puzzle piece, you can create an even smaller collection of elements, which you should also try to make M.E.C.E. within themselves. A puzzle within a puzzle.

    MECEness is a big part of my day. I use it to create product taxonomies, to write better and cleaner emails, to point out problems and possible solutions, to structure excel’s model variables. A big part of my value as a founder and CEO is to be structured. And the biggest part of it is the M.E.C.E. effort.

    Issue trees Sun, 10 Apr 2016 16:09:08 +0000 Issue Trees are a pretty good way to find the solution to a problem. These trees can be more or less exhaustive, however it is important that they are mutually exclusive, collectively exhaustive (MECE) at every step of the way.

    For example, when you are starting a company, you can approach your business plan with an issue tree. In this case, the issue tree will touch a few basic points which the investors will want covered.

    1. What are we offering?
      • What customer problem are we trying to solve?
        • What is the specific pain we are trying to solve?
        • What potential customer base has this pain?
        • What are the differences in serving different types of customers?
        • What is the makeup of the segment to explore first?
      • What breadth of offerings should there be?
        • In a world without any constraints, which are the ideal solutions?
        • In the real-world, what can we offer to solve the existing pain?
        • What are the friction points between our vs. ideal solution?
        • How simply can we convey the value of our solution?
    2. Is this an attractive market?
      • Who is the competition and alternatives?
        • What is currently available in the market?
        • How fully do they solve the customer pain?
        • How do they sell it?
      • What are the key success factors in this market?
        • Which people and skills are essential?
        • Which processes and tools are essential?
        • What is the ideal position in the ecosystem?
      • What are the risks with our approach?
        • What risks are born from suppliers, partners and clients?
        • What legal or other risks and concerns are there?
        • How will conditions evolve and how will we monitor them to stay ahead?
    3. Is this a potential profitable market?
      • What are the expected startup, fixed, and variable costs?
        • What are the costs of starting?
        • What are the fixed costs?
        • What are the variable costs?
      • What is the expected pricing?
        • What is the pricing of current options?
        • How do different pricing approaches compare?
        • What margin can we extract from expected price and costs?

    Issue Trees are also the bread and butter of consultants, as they help them find problems and streamline approaches. But also because doing it this way helps to build a coherent message to the client's top management, and help them understand if indeed all things got covered.

    For entrepreneurs Fri, 19 Feb 2016 00:00:00 +0000 I sometimes get asked by entrepreneurs and colleagues “where do i start?”. This is my small effort to compile a small and honest tool-kit of things I try to use.

    I will start with some simple observations.

    • There is a world of theories, frameworks and tools about how to manage companies. Management can be taught and learned, but most of it involves common sense. So a big part of managing has to come from you being reasonable and sensible just like with any other part of your life. Like waking up super early or maintaining a nice sports schedule.
    • In this sense, management is a big departure from more formal sciences as logic and mathematics, physics, chemistry, etc. It requires a complete difference skill set than to be a great engineer or a developer. Management sits closer to social sciences as psychology - or economics - than other functions in a company. You don’t have to be good at management, especially If you are extremely good at some specific skill, say being a coder or a singer. In that case managers will pile up to do all the 'dirty' work for you. These tips are not for the technical genius, but for the willing labourer who partners with him or her.
    • People who study management before they become top managers slightly over-perform those who did not get such education 1. Still, many of the very best managers learn on the job. Like Jobs. A management education is not absolutely required, but a very nice to have.
    • Great management is filled with simple clichés. Clichés are easy to say, hard to internalise and extremely hard to execute to the necessary level. My top one is “stay focused”. If you want something which is common to all great managers, being focused is it. Stay focused.

    Next, lets describe the prototype manager. What is he or she made of?

    • A Feedback mechanism. You will fail many times, so your first skill is to to be aware of what’s going on and change it fast if need be.
    • Decision-making tools and processss. Vision, strategy, business plans, tactics, campaigns need to be created and put in practice.
    • Leadership skills. You will have to delegate and trust others. Be a resource to the team. Hire people better than you. You have to be a leader.

    This is why my texts are about these categories: Feedback, Decisions and Leadership. There's an extra category about Ideas and thoughts at the end.

    Must-have complex Mon, 31 Aug 2015 00:00:00 +0000 In a startup, at times you'll feel more confident about your product, at times less confident. Don't let that create a must-have complex in you. What is this complex? It's when you receive advice from a VC, a colleague, someone in the industry or an employee on things your business is missing to conquer the world, a task list to success. There is generally the wishful thinking that a combination of this or that trick or tool will solve your problems.

    I've heard plenty of this before, in a variety of arguments.

    The missed opportunities

    • "you need to do more SEO: more pages, link building, full indexing. You'll get many more customers"
    • "we really need a loyalty scheme for our best customers, it makes them buy more"
    • "we received a proposal from company X, they are interested in cross-promoting. Why not do it?"

    Not all good things can be implemented. There's a cost in everything, time and opportunity. But when you're doing something new, you're not improving what you were already doing. Most of the so called "miss opportunities" are things you should explore when you have a big team and lots of money. You want proven, scalable, money-for-value channels of growth. Avoid distractions and things that hurt your product image or brand.

    The customer asked for it

    • "customer X wants to pre-order/ more delivery options/ custom wrapping/ x. We should do it"
    • "we should provide bulk services for B2B, customers asked if we also have a corporate plan"
    • "customer X won't buy if we don't adjust pricing. They're big. Big customers will convert only if we give them a big discount. Let's do it"

    Customers want everything. If you ask them, they'll want the world for a dime. But if your product is good, they'll pay fair value for a problem-solving tool. Separate nice to haves from things that affect every customer from a value perspective. Build and price your product for a customer segment, not 100.

    The deal maker

    • "if only we can make the first order a magical experience! the first order should include a gift and a personal note, that will make a world of difference"
    • "when we launch the new website, customers will flock. This old thing is what's getting in the way of customers purchasing"
    • "they didn't invest, but they will for sure once we launch another city/ release mobile apps/ get featured on TechCrunch"

    Some big leaps are possible and necessary, but all that's shinny and new quickly falls out of style. Both customers and investors can see through it. A great service always treats customers well. Invest equally in old and new customers, and in an ongoing improvement of your product. Focus on user activity, satisfaction, purchase patterns.

    The magic software

    • "can we install that latest analytics thing - GoogleAnalytics, KissMetrics, Adjust, MixPanel, Facebook Pixel, Facebook API? Without it, we won't be capable of measuring x! We'll be in the dark"
    • "we must have a chat for customers to talk with us, it's much more efficient than calls"
    • "software X is a bargain. It does all we need and even provides an API"

    My general feeling is that ~80% of the software I tried and integrated in a startup, we ended up not using. Most likely a 3rd party tool will evolve to match the bulk of their customers needs, which are not always your needs, therefore diverging from your platform. Also, in general, tools don't solve problems, people and processes do. Before embarking on a new software integration, check if it matches your development plan, if it improves any of the major customer pains. Also ask your team if you really, really need it. Finally before you invest time in customizing the so called "easy tool", try to replicate the process manually and check if it's really worth it.

    We can't do it in house

    • "without hiring a PR agency, your business won't get off the ground"
    • "we need a professional photographer"
    • "we need someone with a hardcore CV to get investors in"

    If businesses became successful because of seasoned, experienced professionals, there would be no startups like we know them. Starting a business means you have enough passion to try and solve a problem, independently of the resources (technical, financial) you possess. Of course, having professional help is great and welcome. But it's not absolutely necessary to get your business off the ground!

    Overall, advice that fits this must-haves criteria will invariably focus on doing something new, additionally, like installing a new tool or adding a new feature or doing some marketing thing. More, more, more. Generally, what improves your business is focusing on doing what you already do, but better.

    Growth through complexity is not the best strategy. You don't need to do more tricks or have more tools to succeed, you need to do better of fewer things. Better, better, better.

    Startup Culture Sat, 29 Aug 2015 00:00:00 +0000 A great startup culture is super important, and extremely useful to get everyone motivated. It's not uncommon to hear that a specific startup's culture to be a key factor for success. Zappos, Amazon, Google, Apple, Uber, each have started and grew on the heels of powerful, sometimes bruising, cultures [1][2]. In all of them there are strong figures, outspoken people, struggles, some fights. When you love something, you fight for it!

    But the general flavor and mix, the Culture, is hard to balance. Your designated values already cover main avenues. Some other personal principles pave the rest of the way. Mine are:

    • be there. A big component of successful businesses is actually just showing up. I think this applies even more to the startup culture than anything else. Being there everyday for your team, pulling the hardest, sweating the most, is good for you and good for them. It also means that if you can be there, you shouldn't call. If you can call, you shouldn't email. If you can email, don't wait for your counterpart to email you! Proactivity is a big part of being there
    • breathe startup. If you're in charge of social networks, you should be active in them. If you are in charge of ops, people should feel your energy. Make it a personal priority to show how much you love your area of work. It adds credibility and makes you an advocate for growth
    • be a family. While I don't bring work home and I don't bring my home life into my startup, I consider all my colleagues to be like friends and family. (Reversely, whenever asked, I also treat friends ideas with the same curiosity, structure and rigor as I'd do to any idea at work). I am genuinely interested in colleagues hobbies, where they travel to, what they are passionate about. All this adds incredible color to the workplace. Some people are gym addicts, other are part-time DJs, some are fashionistas or barbecue lovers. No one has to share but it's a hell of a better place if they do, and those things may come in handy, even if it's only at the next summer party
    • quantity matters as much of quality. There will be good days where things go smoothly, and there will be days where opinions will diverge, sometimes in rough conversations. Make sure the nice days are near 99%. A fast environment is many times at odds with doing things in a calm way, and inevitably all hell breaks loose. But people should feel that the absolute majority of their time at your company is great
    • lead by example. If you don't want a behavior to spread, don't do it. Often when I started showing up late for meetings, other team members did the same. The same happens with starting hours, leaving hours, having an active life outside work, checking social networks during the work day, etc. It's hard, and a good example must start in you
    • be personal. In both praise and critiques, I usually address people one to one, and make sure it stays that way. You can make a good example of other people, but be careful to not use only certain people, from certain areas, or praise a limited set of qualities. More importantly, make sure critiques are done directly one to one. I once had a Skype call where I did a very harsh critique of a colleague without realizing he was sitting near other people, which was potentially embarrassing and made him look bad near his colleagues, which shouldn't happen
    • be brutally honest. If someone failed, point that out with a clear example, and as soon as it happens. Having a formal performance review is nice to assess value, commitment and set expectations for a period of time, but it doesn't replace doing it "there and then". If you feel something was a major screwup, tell it like that. If you choose the best people, they'll be able to stomach criticism. Call out bad ideas, and bullshit too. Get technical on your criticisms
    • feelings matter. Even if it isn't your intention, your words can cause other people to think you are against them, or that you don't care. Clarify if it's that's the case. In any case, "politically correct" is mostly a toxic term. But you can be defensive: always re-read your emails before sending, scan for potentially misunderstandings: how may the recipient read this?
    • apologize. You don't lose authority by apologizing, you earn respect. If something hurt someone, even if a mismatch of perceptions caused by your tone, apologizing is the nicest way forward. Maybe you didn't mean it that way, maybe the form or the substance weren't adequate. Maybe the person just can't stomach it. In any case, regardless of the pure business decision of how to move forward, just apologize. It's clean and cozy, but it has to be heartfelt
    • moving along. Sometimes you need to fire people. Or you need someone to change their performance significantly. Sometimes you get really pissed at something. Or you screwed up. Talk about it, learn from it, and move along. There's nothing like a novela to bring your startup to a halt. Be actionable and quick. The team must learn to move along
    • no gossip. "Barracks" conversation is damaging, and rumors can spread easily. If there's something which comes to your attention, before you do something you must know if it's true. Unless you've seen this behavior yourself, or have some pretty good account of it, you shouldn't react to much. "He said she said" is not a nice way to start a conversation. The counterpart of this argument is that people should be open to talking with you at all times, about anything. This prevents gossip from happening in the first place
    • Limit cursing. Cursing is somewhat inevitable in hard environments. Calling something shit is perhaps rough and unclassy, but it's part of the lexico of IT for example. As long as it's not used to describe people, it's a behavior to avoid but which is at this moment woven into the conversations in startups

    [1] The demand of Jeff Bezos

    [2] Some stories on Bill Gates

    Surveys Sun, 26 Jul 2015 00:00:00 +0000 Surveys are a great tool to gauge interest or intention.

    There are transactional surveys, where you ask for ratings of service every time a customer buys. Just like when you finish a ride on Uber. Those should be chosen to match the metrics and KPIs you think appropriate.

    Of particular interest is another type, market research surveys. These are used to understand broad intention to become a user or customer, or to breakdown customer motivation into easier to solve problems.

    The best surveys I've seen are easy to answer, yet they provide answers through inference, not explicit intention.

    When launching our online restaurant service at EatFirst, we wanted to understand the failure of current online food ordering services. On one side, there were the vertically integrated pizza delivery services, on the other you have huge marketplaces where you buy from any type of restaurant. None of them is able to deliver perfect food which tastes and looks as nice as when you go to a restaurant.

    So our objective for a survey was to find if our assumption was correct: that people love restaurants, but it's not always convenient to go out; and that a service that delivers the true restaurant experience anywhere is a great value proposition. Of course, we could just ask those two questions:

    • "do you love restaurants, but somehow it's not always convenient to go out?"
    • "would you be a customer of a service that delivers the true restaurant experience anywhere?"

    But that's making it too obvious to the surveyed people, who may project their wishes too much and end up agreeing with you, and you get little additional information.

    So the team broke down the first assumption in several parts:

    • why do you go to a restaurant?
    • what are your favorite restaurants?
    • which restaurants do you most frequently go to?
    • why, if it happened, are the answers to the previous two questions different?

    With this information you can perfectly validate our first assumption, with details to spare. Eg, you'll end up knowing the reasons why people don't find it convenient to always go to their favorite restaurants (money, distance, time, availability?).

    The same was done for the second assumption:

    • how did you find your favorite restaurants?
    • do you order food online? Which?
    • if it's the case, why aren't your favorite restaurants among the food you order online?

    Now, in this second part, our replacement questions are not directly validating the question. Why? First, a pure, online restaurant, is not something people can instantly think of. It's kind of a novelty, and we don't want interviewed people to get confused. And generally people always say that they're interested in good things, so asking them would just bring very uniform answers and directional, with no measurement of strength. So our objective was to get the appraisal between existing online offerings and the perfect restaurant options and perhaps hint at how to convert them to this new concept.

    Then we built the whole survey, which you can find below.

    Now, executing the survey is another challenge. In this case the whole marketing team participated, me as a founder and head of marketing included. I first wandered in the streets of Berlin at lunch time looking for our target. Almost no one wanted to speak to me, and found only two willing people who seemed to be it (Busy affluent). They weren't. After more than an hour I changed strategy and headed to the subway. I stayed near the track, under the clock with the ETA to the next train. Then I waited for the platform to fill up, and screened for the best candidates. Not only was I able to nail the selection much better, as only one person refused to talk. The fact that I could tell them "it seems like the next train is in 6 minutes, can I use 3 minutes of your time for a super short survey?" really does it. Conclusion: look for relaxed targets in times of idleness. Subway stations are awesome!

    The Actual Survey was the following:

    Greet and explain:

    • this is a survey for a new amazing restaurant
    • its 10 questions, takes 3-4 minutes


    1. What’s your age?
    2. What’s your profession?
    3. Are you single, a couple, a family?
    4. 3 reasons why you go to a restaurant?
    5. Which are your 3 favorite restaurants!
    6. Which 3 restaurants do you go to the most?
    7. If they 5 and 6 are different, why is that?
    8. How did you find your 3 favourite restaurants? (If needed, provide 3 options: personal recommendation, media/ public recommendation, curiosity when passing by)
    9. Do you order food online, from which service and restaurants?
    10. If it's the case, tell me 3 reasons why online food is not among your 3 favorite restaurants?

    Final: Would you like to know about a true restaurant that will let you order online?

    Engineering manager Sun, 19 Jul 2015 00:00:00 +0000 This topic was previously about CTOs. For smaller structures with great developers, an EM should be enough. Its also more practical if the founder team isn't very technical., as the EM will devote lots of time to the dev related issues and careers and yet not code, having more time to liaison with product and he business.

    The responsibilities and some questions for an EM position are:

    1. Lead the team of engineers
      • Build a team.
        • how do you hire a great technical developer? How many developers have you hired in the past? Can you name a few great ones? How much does a great developer, engineer X, cost?
        • when and how do you fire a great technical developer? How many have you let go?
      • Help the team self-organize.
        • What is your vision for a tech team? How much freedom do you give devs? Are they responsible for quality? Describe your ideal development cycle, from end to end, including owners and participants.
        • which tools do you use to organize your team? How do you keep communication flowing?
        • what is your preferred policy on X (e.g. remote work, remote engineers, etc.)?
      • Help the team be efficient.
        • What will your day be like? How will you impact the team?
        • What do you think of the velocity of product development? How do you increase velocity and predictability in the team?
        • Which principles of Agile are more sacred to you?
        • Scrum - What to enforce, what to offer, what to suggest?
      • Manage the career development of the team.
        • How do you want to design the careers of the team? How to motivate them and compensate them?
        • How would you react if X?
    2. Own the technical product.
      • Contribute and align on the product vision.
      • Collaborate on planning activities (e.g. roadmaps) with PMs, Tech Leads.
      • Establish SLAs for uptime, availability and performance.
      • Ensure value is continously delivered to users.
        • Do you want to code too?
      • Operate, monitor and maintain product.
      • Establish on-call duty processes.
      • Create appropriate reporting.
      • Manage the teams budget.
    3. Establish yourself as an ambassador.
      • Act as an ambassador to the product, inside and outside organization.
      • Act as an ambassador to the tech team, inside and outside organization.
        • which technologies do you follow more closely? how well known are you in their communities? How close are you to the leader? What are your major software / hardware contributions?
        • how many people follow you? How many people do you follow? Who? Why? How many would try to follow you to our company?
        • which companies and organizations are you affiliated with? What's your relationship with them? Who do you know at Apple/ Google/ Amazon/ X?
    4. Values
      • Act according to the company values. (Depends on the organization).

    Remember it's extremely important to triple check all recommendations, and you should get a bunch.

    Marketing managers Sun, 12 Jul 2015 00:00:00 +0000 Marketing manager is a position that includes everyone from a CMO to a sales operator. All should have extensive sales experience and should be in permanent contact with your target customer. After all, they are the first contact with the costumer. They bring home the bacon. Whomever wants to be in marketing must love a product and be ruthless and objectives oriented.

    Again, I check them for their 3 responsibilities:

    1. Sales, sales, sales

    • how much do you think we're selling? How much do you think we could sell? If I hire you, how much will sales grow in 1, 3 and 12 months?
    • what was your biggest sale ever? Tell me all about it
    • how many potential customers do you know? How many can you convert within 1 week? How would you find more?
    • what are the 4 Ps of marketing? what's important about them?
    • what do you think of our product/ service? what are our products' key selling points? How do you assess it versus the competition?
    • talk to me about our pricing strategy
    • how well are we placed? Do you know these channels well? How would you characterize them? Tell me about performance and non-performance channels. Online and offline. What is the reach, the key influencers, the customer education needed for each?
    • what do we need to do to get a customer on board? What is the sales funnel? How much incentive? How do you keep them?

    2. Create quality campaigns

    • how do you design a campaign? What's the starting point?
    • tell me all about CTAs
    • how do you measure campaigns? What are the best KPIs to track marketing performance for our company?
    • what are leading and lagging indicators?
    • how do you make campaigns better? Does AB testing really work?
    • design a process to track a whole team of sales people - for marketing, for a particular channel, or campaign

    3. Keep customers excited

    • create a few variations of the X CTA
    • why do customers stop buying?
    • what qualifies as "customer amazement"?
    • what is the prettiest thing you've ever done?

    At the end ask for the relevant references and cross check them with the story which formed on your head!

    Business Intelligence manager Sun, 12 Jul 2015 00:00:00 +0000 The BI manager is an extremely special role. Generally people who support decision making are necessarily part of the operation. They are very connected, on the ground. They have daily ownership. But not the BI manager. The BI manager raises the bar of intelligence throughout the company, wherever she or he is called.

    To me, a BI manager has 3 main responsibilities, and these are the questions that trigger my understanding of him.

    1. Create impact through BI
      • What are we doing in this company?
      • What's the objective of this company?
      • How would you measure this success?
      • How do you classify different indicators? Tell to me about leading and lagging indicators.
      • What is performance? What is productivity? What is output?
      • How do you track performance? How do you predict performance?
      • Tell me about the biggest impact you've created in a company. How do you know it was you who caused it?
    2. Develop informed recommendations
      • how do you organize yourself?
      • how should people ask for a new study from you?
      • what are your favourite tools for analysis?
      • what are cohorts? Build a cohort model
      • what is SQL? Create a simple query for X
      • how do you project an operation into the future?
      • (provide file) what's going on in our business?
      • Tell me the last time you screwed up some data. Why did that happen?
      • What is an attribution model?
    3. Hold a standard of Experimentation & Truth
      • How do you build appropriate AB testing?
      • Dali didn't have BI. Coca-cola and Hermes didn't have BI when they started. Is BI really that important? How?
      • How do you vote?
      • Tell me about astrology
      • How should I evaluate and compensate you?

    Don't forget to cross check his or her answers with external references.

    Product managers Sun, 05 Jul 2015 00:00:00 +0000 A good product manager or product owner carriers the idea of the company product forward. A PM owns the roadmap and works on lining ideas, challenges and problems. PMs balance the questions: customer purported needs, UX explorations and tech teams potential solutions. They provide a personal view but draw mostly from others. They spec and draw and speak the tech language.

    Apart from specific company fit, my questions to candidates are about the 3 responsibilities of a Product Manager:

    1. Improve your business through a better product
    • what do you know about our company? What are we doing here, what's our potential?
    • what's the impact of our business model in designing a Product? What are the critical challenges? Do you have the skills and experience when dealing with similar problems? Explain.
    • Where is the industry going?
    • What's impact to you? List the 5 most important ways in which a product improves. How would you define success in your job? Tell me about your biggest impact in a product management role you've had. How did you measure that?
    • Do you have experience in working with a team like ours? What do you think about X as a technology?
    1. Adequately balance scope, time, cost and quality (Project Management)
    • What is an effective project management? What are the variables to optimize?
    • How do you organize yourself? How do you communicate with people? How do you detail an issue?
    • Talk to me about Agile, Kanban, Scrum, X. Which do you like the most?
    • In 60 seconds, make an accurate picture of X? Do you know what MECE is?
    • What do you hate about our product/ another product? Why? How would you make it better?
    • What features would you kill in our product/ product X? Prioritize them for killing until there is no product left. Where would you draw the line of what gets killed and not? Why's that?
    1. The "one more thing"
    • do you think our product is beautiful? What's magic about it?
    • tell me about the most incredible experience you've created
    • how do you systematically create those?

    And that's all! Now, people aren't required to nail all of the questions? I wouldn't nail all of them, for sure. Between Values and these Technical questions, a good interview will take you perhaps some 2 hours. Some fare better in some questions, some in others. Cross check the most obvious answers with their past managers. Some questions don't have a right answer. What matters is that you get a good feeling about who you're about to hire!


    [1] Deconstructing hiring through values

    About Mon, 04 May 2015 00:00:00 +0000 > Berto <

    My name is Humberto Bento Ayres Pereira. I'm from Porto, Portugal. I currently live in Berlin, Germany. I build products. I have a wonderful family.

    I am building a life together with the person who makes me the happiest. We have a daughter. We also have our mascot dog, Lopes.

    I also have a serious play life which my family fully supports:

    2016-xxxx: Founder of XXX, a next-generation productivity tool.
    2014-xxxx: Investor at, a sports booking service. Profitable. 2014-2016: Founder & Co-CEO at, an online restaurant.
    2012-2014: Founder CEO at, a Health & Beauty biz. Now profitable.
    2009-2011: MBA at Harvard Business School.
    2008-2009: Consultant at McKinsey in strategy & tech projects.
    2006-2007: Caltech road vision research fellow for the DARPA race.
    2002-2007: EE+CS Engineering at the University of Porto.

    I enjoy reading. I really like cycling - no tight shorts, more like cruising.. I also really like running - occasionally with the tight shorts.

    Law and contracts Sun, 19 Apr 2015 00:00:00 +0000 I consider myself a lucky person. I have found myself in situations with ethical grey areas, and I was fortunate that there was someone more experienced (or more conservative) to guide me to do the right thing and save me from potential hassle. I tend to trust my impulses and intuition as much as my purely rational judgement. I always want to start fast, too fast, and consider contracts and accounting a chore. I have stayed out of trouble despite my over excitement and impulsiveness. This has not always been the case and I have personally and financially suffered.

    I learned my lessons though. Now, whenever I am making a big decision that has other important stakeholders, I always try to consider the angle of the Law, in its practical and purpose (ethical) sense.

    One episode from my past Is a good example of what not to do.

    • when I started this venture, me and my angel partners signed an agreement. I had every reason to trust them. They were close and were investing a relatively big sum on essentially a blank slate. Out of respect, friendship and perhaps some embarrassment, I didn't get contracts thoroughly reviewed by a lawyer of mine. I asked all the questions to their own lawyer, sent the contract around to friends for quick comments, and we moved on.
    • after 2 years in, we shareholders had diverged in what strategy to follow for the future, and I had resigned amicably, keeping my shares. Six months later we had a disagreement over how to raise money to keep the company going with enough comfort. The company was operationally profitable in the month I left, and was growing rapidly. But the company had also amassed some debt with suppliers, before and since my departure.
    • As an anonymous partnership, we were liable (mostly) only to the State and Employees. But some clauses could allow a different interpretation. The angels wanted me to contribute capital and also sign a personal guarantee for liabilities covering the time I had been a CEO but also later times. I didn't want that. I suggested they put in more capital and dilute me, they didn't want that. To make matters more complex, my partners were also majority shareholders of our main supplier, and therefore the ones to benefit from any immediate payments and guarantees.
    • As each side tried to explain their positions, it became very clear we weren't going to agree. Our Z.O.P.A. hardly existed. I proposed that we elect an outsider committee to settle the issue, but that was rejected. The tone soured and egos inflated on both sides. Discussion turned to grandstanding.
    • From an intellectual perspective, I loved this situation. I had done nothing wrong, and I had always worked as hard and professionally as it is humanly possible for me. I thought nothing wrong could come to me. Plus, I was super curious! But time was going on, and a solution was needed.
    • I proposed covering my share of the maximum potential liability at the time of my exit - including all unsold inventory at the time of my departure and all supplier credit at the time too me, which is a stretch. This was small money but personally at the time it was a lot for me.

    All of this would have been avoided if I had involved a lawyer in the beginning. It was imprudent. The resulting contract could have helped everyone manage this situation better, saving all the other shareholders and myself a lot of headaches and money.

    A clearer, simpler contract would have limited the pressure on both sides of the argument and reduced complexity. Starting a legal discussion is the same as starting a real war: It's better if everyone's holding pen and paper, not sophisticated and lethal ammunition such has money. The contract was not 100% clear on a startup character, rather it resembled a "marriage until last consequences", and it was my fault for signing without checking it through.

    As a result, whenever I am about to enter in a contract, I hire a lawyer. I ask all the questions, both stupid and smart. He'll be forecasting the better and the worse, and even more important, the way laws can be used in different scenarios.

    Trust and optimism is fundamental, but it's smart to exercise your right for protection. When you're at the negotiation table, save a seat for the Law.

    Cohorts Wed, 08 Apr 2015 00:00:00 +0000 Whenever you launch a new marketing campaign, or a new product line, or get some good old PR, you want to see what’s the effect of that action. Cause and effect. The easiest way of doing it is to track cohorts.

    Cohorts are just groups of people that share a common characteristic. As your business progresses with time, the easiest place to start is tracking time cohorts. So a cohort would be defined as the group of people who first bought in a certain day or week or month. How many customers did we acquire this week vs. last week? How good of a customer is a customer who bought on christmas vs. one which bought on valentines? All these questions are answered if you track your cohorts.

    The simplest time cohort has two axis. On one axis you have the date of FIRST purchase of the customer. On the other, you have the date of ANY order. Just like in the picture above, you see that the diagonal is the first time a customer purchased — customers who first purchased in march who also purchase in march are first time customers. Also, for each group of customers who FIRST bought on a specific date, ANY purchase with date bigger than REAL date is a return purchase. Simple.

    How to build such cohorts?

    1. you start with a list of all orders. The minimum information is a customer id and date of purchase. Believe me, with that you can already build something super powerful. If you want to do other things, you can include spend, number of items on the basket and other information.
    2. then you need to mark, for each order, the date of that customers’ first purchase on a new column. For first purchases, the date of that order and the date of customer’s first purchase will be the same. All you need is a vlookup.
    3. then you can build a model yourself (eheheh) or just throw it all under a pivot to crunch.

    You can check and download the following file with the individual steps here: my first cohort table. You won’t have permissions to change anything, but you can download the model by going File/ Download/ Excel.

    On this particular example, if you look at step 3, we can see that the first month had a lot of activity (27 purchases) and that on the following months we some mild activity (between 3 and 6 purchases). The next 3 months cohorts only had 1 person each, and this three people (1 per cohort) came back actually quite a lot comparatively (between 0 and 3 times on each successive months). So these cohorts of people show a better behaviour relative to their initial month of purchase than the first cohort, which was big (27) but didn't have a lot of returning action.

    With a cohort like this, you can:

    • understand if you are acquiring more or less customers per period
    • understand if each new cohort of people is returning more or less often
    • extrapolate a natural timely behaviour of people and do projections
    • compare all this with real world actions (campaigns, product changes) and understand the impact of your actions

    So, what other types of cohorts are there?

    • time cohorts. The ones i described above
    • campaign cohorts. Instead of grouping customers by date or period, you group them by the campaign which acquired them, independently of when they were acquired (or in addition to)
    • behavioural cohorts. Really, any trait can be used and analysed. How do female customers behave in the months following a purchase vs. male customers?

    The data plotted in a cohorts analysis can be varied as well:

    • orders. How many orders each cohort created on their first and subsequent months/ period
    • revenue. How much revenue each cohort brought on their first and subsequent purchases
    • items per order. Yep
    • fancy something kinky? Really, anything related to the orders can be crunched

    Finally, there’s an interesting twist I like to give to cohort analysis, which is if you do your analysis orthogonally. Orthogonal means things are independent from each other. So, in a way, instead of doing a big cohort which tracks revenue, you can build several cohorts that track users, orders per user and revenue per order. Each of these sub-cohorts can be affected by different events:

    • boosting your customer acquisition campaigns will affect mostly the amount of users
    • opening on weekend probably affect the amount of new customers and orders per user
    • launching a companion product will most likely affect revenue per order more than anything

    Hands-on time! Check out this cohort model that you can explore. It tracks several orthogonal variables across the real data you input. It tracks their evolution and their projects the growth of current and future cohorts from those variables. Finally it uses a convolution of all the cohorts to project revenue.

    In order for anyone to access it, I put it on Google Drive Orthogonal cohort model. You won’t have permissions to change anything, but you can download the model by going File/ Download/ Excel.

    Have fun!

    Be your own boss Tue, 24 Mar 2015 00:00:00 +0000 I have heard many times from colleagues and friends that "life's just about enough". I've never truly believed in this. Everyone feels special. Everyone is special. And very likely everyone wants to continue special.

    And so in my view there's not one person who is not ambitious, it just happens that there are many kinds of ambition. Some want to be artists, other politicians, some want to be great lovers, others rock stars.

    Some want to build something for the world. Yet it's hard to start at the top. Many get impatient at their stepping-stone jobs. They want to be at the top from the start. They want to be the boss. Fast.

    Well, it's easier than you may think. All you need is to start by being your own boss.

    How to achieve this even when you're not in fact a boss?

    • A boss has objectives. Ask your manager for yours, challenge them and push yourself. Always make yourselve chase goals that make you uncomfortable. Start by your own personal goals, like how soon you want to wake up everyday. Try to wake up very early. You'll never be a great leader until you can control yourself, so start there.
    • A boss has a budget. Managing money is essential. Talk to your manager and ask for yours. How much money can you spend on your function? If it's not money, how many hours? How many people can you manage? How much stock can you spend for your purposes? How much revenue is expected of you? Or what part of the revenue target is attributable to you? Operate your day like it was a profit center, managing revenue and cost. Good value is always expected from investing in you and your tools, the more you make this evident, the stronger you will push and the faster you will see recognition.
    • A boss closes the loop. Customers behave in a cycle, you can't just acquire and sit still; chase them and learn from them. Don't consider a few competitors, track and learn from all substitutes. Don't be narrow minded, your metrics should cover performance all the way up to outcomes. Don't create to-dos of things you won't follow through. Study the past, act on the present and look out for what's coming. In other words, simplify but never forget the whole picture. Be MECE. Close the loop.

    If your plan is to one day be a leader, you have to avoid at all costs being isolated. If you are not on top of all aspects of your trade - directly or indirectly thinking about strategic, organizational and operational - someone else will have to be. And you'll miss out on the opportunity to grow.

    Improve on all areas. The most important thing is to not give yourself some slack. Act like your area is your own mini-company. Be ambitious like it's your life. Spend like it's your money and time. Don't miss out on detail.

    Be your own boss.

    Leading from the bottom of the org chart Sun, 22 Mar 2015 00:00:00 +0000 HR-wise, the place of a great leader is at the bottom of an organization.

    A great leader at the bottom:

    • recognizes that the companies many employees are the front line of the business's and gives them corresponding responsibilities and credit.
    • hires people better than himself or herself to do each value-add function.
    • maintains a super high level of motivation and performance by providing accurate, objective goals and feedback.
    • recognizes that the job fit depends on the person, the function and the specific present objectives of the company. It is not personal, but sometimes certain employees need to go.
    • acts as an internal consultant, not vested in any scared cows, who only wants to help making better decisions. The consultant leader fights for structure, not for a particular solution. The consultant leader is fair and balanced. The consultant leader helps iterate the current company into a future vision.

    I personally have a hard time not jumping into every discussion and Leading from the bottom. I tend to micromanage at the slightest doubt that someone may not be on top of his or her game. It's a work in progress for me, though I think I have made some progress. I now surround myself of people who are more experienced, and act much quicker when I don't see that, by changing roles and people.

    So make yourself the bottom of the organization. Because a company culture where the CEO and other managers are there to serve and help the team is a great one.

    Pivots Thu, 19 Mar 2015 00:00:00 +0000 The opportunity and risk of changing something in your startup is a daily matter on the head of every manager. It's hard to decide what to do and what not to do. Especially when you've already invested a bit in the status quo. Product, system, processes, team, brand. These past decisions created company elements that provide some flexibility, but maybe not all you need.

    Any decision can be the best thing or a catastrophe. But you should look at pivots, big and small, in a positive light. Opportunity and risk should indeed scare the shit out of you - but it was your decision to join the startup world in the first place. The biggest risk was joining and that's already past! So be happy about the flexibility.

    My general rule of thumb is to implement all ideas that improve whatever the customer is already a happy buyer of. Everything which you are currently NOT providing him and that you think that the customer wishes is dangerous. More often than not, you should not do it.

    I consider worthwhile exceptions the cases when you're currently not doing something but there is significant overlap between this new thing and your current thing in all categories:

    • Who. Same Decider, Payer, User, Consumer?
    • What. Same product category, brand, price points?
    • When. Same Time of sale, Repurchase pattern and Sales funnel?
    • Where. Same Target location?
    • How. Same operations for sourcing, production, marketing, sale, delivery and customer service?
    • Why. Same pain solved?

    If all or many of these 18 items in the 6 categories are the same, then it's probably just an extension, not a true pivot. If you're changing most of them significantly, then you may as well close the company and setup a new one.

    Another aspect to consider is wether the current offering is performing well or not. If it is, you're considering adding features or a new business line which may improve business; risk is small if you are responsive and scale the changes smartly. If the current offering is not good and your cohorts are miserable, your pivot can be a life boat; or pivoting can be just running away from one problem and looking for another one.

    In any case, you should consider the complexities of the iteration. Will you end up with 2 teams or one. Is investment big or small. How much cash do you have and how the team will take it.

    One clear example is the move from B2C to B2B or the reverse. While the product may be similar, or technically the same, it’s not an easy pivot. Its tough because you change many things in Who (Decider, Payer at least), Why, a bit of When (Sales Cycle) and How (Marketing, Sale, Customer Service) and possibly What, Where.

    Directions change. But keep in mind that there are plenty more roads to doom than to success, and sellers of those strategies are plenty and close to the heart. Follow your own vision, informed by the needs of the customers, the direction of the industry, data and optimism. It is the best proposition. The important thing is to have an objective discussion around what it takes to pivot. It will still be harder as there are many unknowns, but at least you can chart the effort. And in the end, follow your will.

    Leading and lagging indicators Sun, 08 Mar 2015 00:00:00 +0000 The first part of being smart is to be aware. To be aware of what's wrong and what's right and how you're progressing. However, in the past I have been completely fooled by my flawed perception of what to be aware about. I have also found that I wasn't the only one. So I'll try to go over the most basic element of BI, indicators, and how you can structure them.

    Take a car manufacturer. For the business person, it's important to know sales numbers, revenue, and overall customer loyalty. But to tell those numbers to a factory worker who is actually building the car can be unproductive. Yes, the quality of the cars affects the satisfaction of a customer, and thus impacts sales and consequently revenue. However, tracking that for the worker has little value in daily operations. Each car will take at least a few days to be finished and sold, and at least a month to show up on sales reports. Also, there are many other people involved in selling a car and providing assistance besides the worker. It will be hard, then, for the factory worker to know how he's doing based on customer loyalty - also because there are many more variables at play. It's much more important to know how fast he finished his tasks, or how many faults his work contained.

    So how should one approach indicators? As in the example above, we see that faults in the car production is a very immediate indicator, whereas loyalty takes some time to be affected by the workers' impact. We say that faults is a leading indicator of this workers impact and that loyalty ratings are a lagging indicator.

    Dividing indicators into leading and lagging for each function to be analyzed is to me a crucial task of Business Intelligence.

    Leading Indicators characteristics

    • They try to measure Performance. These indicators signal future events.
    • They are inspectable. Leading indicators usually refer to a baseline, or a standard performance against which we are measuring the subject.
    • They are individual. This means that they are individually actionable, making them fair and less political to act on.
    • They are instantaneous. This makes them easily observable and quick to diagnose and correct.
    • They are immersive. They reflect the whole of the responsibility of a function or person in some detail.
    • They are independent. Each leading indicator should be as much as possible unaffected by other indicators. People should in principle be able to optimize each of them separately.

    Example of leading indicators for the performance of our car factory: speed at which the worker picks his screws to be bolted on to the car. Speed at which which he bolts each of them. Number of retries due to error. Number of screws needing fixing due to low quality work. Number of break pauses and time spent in break pauses. Again, these should be relative to a standard measure. Leading indicators mean that they improve before the system as a whole improves.

    Lagging Indicators have the opposite set of characteristics

    • They measure Outcome (Output). These types of indicators follow past events.
    • They are objectives-based, aggregate, cumulative, high-level and complex.

    Example of lagging indicators in our car factory are number of cars produced, average total cost per car, customer satisfaction with cars, revenue for the factory (in case it's tracked). Lagging indicators change much after the company has internally improved.

    In between Leading and Lagging indicators we can establish a number of levels of lag or lead. Typically this concerns not Performance nor Outcome, but rather Productivity. Examples for the factory workers could be cars per hour, accidents per month, and throughput.

    To make an analogy, we can compare your company to the movement of an object. A force is applied producing an acceleration. Both force and acceleration are Leading Indicators to the performance of the mover towards the movement. Between them, force leads and acceleration lags because acceleration is produced by the result of possibly many forces. Speed and mileage lag behind as indicators, and they measure productivity of the mover towards the movement (km/h) and (km/ L). Finally, dislocation, total time or total costed the trip measure the output, and are indeed lagging indicators towards all other indicators.

    Both types of indicators are very important. Leading indicators are good to establish the root causes for improvement, and lagging indicators are generally what the investors want to hear. The most frequent mistake is to evaluate workers by their productivity (pieces produced per hour), when many times this process is not entirely under his control. Maybe the worker has to wait for parts to arrive, or part feeding varies greatly according to personel available and hour of the day.

    It takes time to build a fair Indicator system that can motivate people into doing a better job, but it's absolutely necessary if you want to consistently improve your business. Keep your leading and lagging indicators in check for an optimal and fair process.

    Writing shorter Wed, 04 Mar 2015 00:00:00 +0000 Long emails, long sentences, long excuses and long explanations. Eventually your audience will get bored of them. I have in the past been extremely guilty of that. Partly, I'd say, because of the traditional writing culture in Portugal. There is an abundance of asides and inclusions in speech with parentheses and commas.

    But it's a bad habit.

    How short is short enough? Apparently, anything larger than 7 chunks of information at a time is too much. And so, each sentence, paragraph, list and rationale should be kept at a maximum of 7 chunks of information. Probably its better to keep it at lower quantities still.

    How to eat an Elephant? Wed, 25 Feb 2015 00:00:00 +0000 One slice at a time. The oldest adage in Consulting is perhaps one of the most valuable insights for startups.

    People think that scale is all good. That the bigger you get, the better you are. This is not necessarily true. I've seen many companies get lost when they try to tackle too big of a problem all in one go. Governments projects are always under budgeted (time and cost) because they are too ambitious and improperly planned.

    Complex problems have many dangers.

    At the top is Demotivation, because you get stuck in it for a long time and you never get the satisfaction of completion. This can be super bad for the morale of your colleagues. This is close to the book writing problem. When you start you are very excited, but a few thousand words are not enough. After many months, it can be tiresome to always go back and see the same unfinished idea.

    It is followed by Inefficiency, because the more people and resources required to solve a problem, the more time you will spend in coordination and synchronization, and so the more cost and delays it is prone to have. Because of this, enterprise solutions frequently have marginally better features at a disproportionately bigger cost.

    On top of Demotivation and Inefficiency, there's Obsolescence, because when you end it, the initial conditions may have changed. This is also why when most big software gets released, it already looks old.

    Finally Dependency is super bad because by pursuing a big project you become too invested in that particular solution for a problem. It can very easily become too late to go back, and yet still too far away to complete. This "all or nothing" can be very costly.

    So take your time. Eat the Elephant piece by piece, be it a software product, a new business line, or a brand restart

    You'll be more motivated, more efficient and will have a fresher result to show.

    The unknown known Wed, 25 Feb 2015 00:00:00 +0000 A consequence of society's accelerating evolution is that people feel out of place.  Example: Out of place in the world, because they need and want to travel more, yet everyone likes to feel "at home" wherever they end up.

    Why is so difficult when we never had access to so much data and tools?

    Precisely because of the abundance in data, I'd say. The traditional top down approach to extracting value out of data is by organization and filtering. But that's too slow and expensive, and too narrow scoped. So, hierarchies got replaced by networks [1]. For example, Foursquare is not a city guide, instead it relies on the input from thousands of businesses and users who register places and checkin. They make this to avoid creating an city guide which is good for me but bad for you. It also allows them to simultaneously launch in many cities and Countries. This makes the network more resilient, complete, and fun. However, not all is rosy in a Place network. These networks share connections around a common discipline, and try hard to be personalized but they are too commercial. And commercial, like cheap music selections of hits on the radio, don't feel personal, even if you like them.

    Too much content is simply too hard to navigate. Sure, Foursquare tells you places your friends check-in, and suggests nearby places, but somehow that doesn't seem enough. Why would a place someone I know visited be good? I can train an algorithm to try to infer what I like from other users and categories, but so far the results are too bad.

    Let's see a practical situation. Everywhere I go, I always try to find where:

    • to get a native Japanese meal as good as in Sakura in London (now closed)
    • to run in a scene as natural and grandiose as the marginal of Porto
    • to eat a smooth stroganoff like in Gorky Park in Berlin
    • to listen to nicely jockeyed old tunes as in ZuZu of Boston
    • to party in a discotheque as cool and customer friendly as Lux in Lisbon
    • to find the nearest rooftop that rivals The Standard in LA

    I do search Foursquare and TripAdvisor and the like, but they are pretty much useless for this. Consequently, when I go to a city I mostly ask a local to guide me, and generally I am happy with the results. But its suboptimal.

    I've been thinking about Explicit category networks. Say, I like speakeasies in NY. Particularly, I like Apotheke in Chinatown. Now, I'm pretty sure there's someone out there who can tell me what's the closest to Chinatown in London and Berlin. Or a bar which provides he same breed of live music, great cocktails in a private environment in Paris. This way, I would be able to experience the local feel of a neighborhood, park, bar, restaurant or library but In a way that would kind of be familiar to myself.

    These services would still be networks, because they (could) rely on an a multitude of opinions, yet somehow hierarchical in the way that it would be restricted to a certain domain and force, not simple signals.

    "When in ____ I felt like I was in ____" would be the currency of this service.

    I'd likely and gladly become a customer of a service which solved this problem.


    1. Wikipedia is not edited by a fixed body of appointed experts. Small projects are Kickstarter'ed not by big companies but rather a list of people with the same interest. Bitcoins are "regulated" by a headless body of developers and miners.
    Internet, the toilet paper of the 21st Century Wed, 25 Feb 2015 00:00:00 +0000 The internet has the potential to be the most democratic, revolutionizing tool Humanity has seen. It is kind of fulfilling this promise. But along the way it's become the world's toilet paper too.

    I consider a characteristic of Humans not to have smart things, but to be smart themselves. And we need to improve to be considered smart. But most end-user web services are useless at helping us precisely because they fail to distinguish between Being and Knowing. While they are smart under the hood, they don't provide a way for the customer to Know and appreciate that, and with this knowledge improve themselves.

    Take Facebook. Your News Feed scrolls down and down, infinitely, always showing more info. You do a lot of interactions with the platform every year. But there's actually little to learn from your feed and your account. For you at least. While Facebook sucks in your info and has amazing algorithms to choose what you may like or not, they don't make this choice, learnings or its mechanics visible to users.  Facebook is giving some small steps, because now they actually present you with the highlights of your past once per year. So in reality Facebook is not toilet paper. It's double-sheet toilet paper.

    On to Gmail. Dozens of thousands of emails back and forth, and what do they do with all that context? Ads. They could help me improve my writing. They could map my friends and tell me if I'm not contacting them enough. They could improve the success of my emails [1]. But no. They get all the fun. Gmail serves as another support where we smear our communication history.

    Twitter, Instagram, Pinterest, all of them have this same characteristic.. They are a medium to your shit, and then it's all flush down the toilet into obscure dirty pipes.

    So, I've put together a list of similarities of toilet paper and most internet services:

    • everyone uses both of them now; hard to believe as it is, in the past no one used either
    • each individual piece of any of them is worth shit; each additional piece is still worth shit to you, no matter how many you put together
    • for some people, putting together your used pieces and everyone else's functions as a fertilizer of their business. They pay you shit for it
    • the real world is about tasty stuff, the version which ends in there is shit
    • both are usually cheap to have
    • you shout when there's an outage of either
    • both are extremely long, but you also consume them pretty fast
    • you use both to store stuff no one really has an interest in seeing

    I do know this view is simplistic, in that it doesn't consider the main use of the mentioned platforms as "useful". Of course, sending emails and communicating with friends is very useful. My point is that while atomically and instantly useful to us, they do not allow an empowerment of the user through memory, logic and visualization.

    What would the non-toilet paper internet look like?

    • a reminder. With things you truly mentioned you wanted to do. Not ads of stuff other people want to sell you.
    • an organizer. With neatly smart tied events for stuff you frequently do. Not a calendar filled with other people's spammy events.
    • an assistant. That goes beyond reminders and organization, and that has an intuition. That knows the best flowers to send and when it's the right hour to ask you "when was the last time you called your grandma"?
    • a shelve of the best selection of photo albums. Not a million pictures in a stupid grid.
    • a story of your life. For yourself. Not to break down in segments for preying advertisers.

    Disclosure: I do use these services and I also frequently advertise in them. Also, this very repository of texts is like toilet paper, though I force myself to review past texts and refer them in more up to date versions.


    1. Please drop everything and go read "Avogadro Corp" right away!
    Product fail Sat, 07 Feb 2015 00:00:00 +0000 Building a Product is about making something so useful, simple and elegant that eventually you stop thinking about it.

    I therefore usually consider 3 main fails in products, when each of these conditions are violated.

    Fail 1. Utility fail

    Things which are not useful. Either because they are just not providing a benefit, or because a benefit is complex to attain. Look at the two following products:

    • Microsoft HoloLens
    • Google Glasses

    While those are good technologies, they currently solve no big problems for any big category of users of the companies. They were announced without a purpose. They are just an enabler technology. If developers don't jump onboard, and Microsoft and Google can't come up with a specific killer application themselves, these devices won't fly [1].

    On the opposite side, there's stuff which does provide enough utility at launch despite obvious trade offs: Tesla cars, the Amazon Kindle, GoPro cameras. A good product doesn't have to be expensive or rule the world. But it needs to provide enough function to solve at least one problem.

    Fail 2. Simplicity fail

    Many things are hard to operate, blocking the products or services growth that could be afforded by their utility.

    Take for example banks. Banks are very useful. They provide you storage of money, credit for your projects and applications for your investments. But their websites and online services are traditionally hard to operate, complex, and assume you know bank jargon, like the difference between Debit and Credit. And I'm not talking about the cards. Some have extremely hard to find features and don't allow you to access some services online - like subscribing to a new card or searching for transactions by the name of the merchant. Worse, they have intricate and bad UIs and very convoluted security features, some of which are going in the opposite direction of the whole tech world.

    I've used plenty of online banking. The worse I've used are, as of Feb 2015:

    • HSBC in the UK. They require a token gizmo which should be replace by better secrets, SMS confirmation, or integration into the smartphone security. The UI also sucks a lot.
    • Deutsche Bank in Germany. With its TAN matrix authentication out of the 19th century, mobileTAN which you have to pay for, and incredibly dysfunctional UI and features, the service is hard to compliment.
    • Caltech and Harvard Credit Unions in the US. Stuck in the 90s, they nonetheless don't make your life as painful as the top 2.

    The best:

    • Bank of America in the US, with nice smartphone integration and overall organization of info.
    • ActivoBank in Portugal, by far the simplest yet capable service. They have a smart expense manager and all, and they have a truly nice mobile offering. (Disclosure: I participated as a consultant in the project which resulted in AB).
    • Simple. Though I have no account in Simple, I've seen it in action and it makes me hopeful about the future of online banking.

    Fail 3. Pleasantness fail

    A much finer mistake is the pleasant fail, for those ideas that provide enough value and are even simple to operate, but that lack a view of the user as a sensible being who likes coherence and harmony. While subtle, this fail is in my view what distinguishes a good performer from a leader product.

    There is no clearer way to demonstrate this than by going to a bathroom. Go there. Now open the water and wash your hands. Close it. Now you're done, answer the question: did you touch the washbasin or the faucet at any time during the hand washing? Probably. Why? Because of a bad product. What I've observed is that most of the bathrooms have the wrong size of the water faucet, and so the water falls too closely to the sink edge, making you occasionally rub your hands against it. Now, you may say that it's not a big deal, or that I'm just another victim of obsessive compulsive behaviour. Probably. But there is no denying that it's not a perfect experience, and that reaching it would not be extremely hard. All they had to do was choose an adequately sized faucet. And the pain may not be big in your house, but if you consider all the airports and shopping malls which have these awkward washbasins which make you rub your hands against the porcelain, then clearly some architect or designer or contractor didn't think hard enough and is to blame for that feeling.

    These flaws create the feeling that something is not as good as it could be. The feeling of an inferior product.

    1. An Microsoft ex-executive sums it up perfectly: "Molyneux commented, "The bizarre thing is a huge amount of effort and time and money goes into researching the tech, like the Kinect tech and scanning the bodies, and there's always this one line that hardware manufacturers - whether it be Microsoft or anyone else - say and that's 'we can't wait to see what happens when it gets into the hands of developers.' Now if Apple had said that when they introduced the iPhone, I don't think we'd ever end up with the iPhone! What really should happen is that they put a similar amount of money into researching just awesome real world applications that you'll really use and that work robustly and smoothly and delightfully."
    Fight for structure Sun, 01 Feb 2015 00:00:00 +0000 "Suddenly, that simple question turned into an email thread with 30 responses and 20 participants. I felt ignored and challenged in what was supposed to be my responsibility, and the whole company was watching. I still thought I was right, but others had valid points too. Most of all, I didn't want to feel dumb, and I also didn't want to come across as an idiot. All I ever wanted was the best for the company."

    Uncertainty spawns dissidence. In a startup environment, things move fast. Not only because the company is small and you have to gain traction in the market before others, but also because the path to success is unknown. The CEO keeps speaking about product-market fit because the company is still not in Auto-growth mode. This uncertainty brings multiple options to the table to every area: do we try flyers or door-to-door sales? Do we enhance a previous version of the product or start anew? Which are the best KPIs? Outsource or in-source? Trial a new business line or keep it all-in on the current idea? What stuff should we kill? Multiple ideas have different backers from Marketing, Ops, IT. Even investors, board members and customers will have different options. At any role, you will be asked about your ideas around a topic, and those will be discussed and evaluated by other departments and in other realms, and others will disagree on your views! Shocker!

    Then, invalidation brings frustration. The more an idea gets discussed by different departments, the more diverse the arguments pro and con are likely to get, arguments of different natures as time, cost, focus and impact. People start discussing all these arguments and keep adding more to support their own idea or rebuff others. Because you've probably spent more time dedicating on what you think is right, you are naturally incentivized to add depth to how beautiful the future is if the company goes down your path. Everyone else does the same. Some people bring rationales, some bring outsider examples, some bring practical cases of your company. As one of the stakeholders you get frustrated, not only your idea is being questioned as also there are no signs that anyone's idea will ever be approved with consensus.

    A way to solve it?

    Start at the end: The best ideas should always win. Basic. So, if you recognize there are plenty of smart people around, you should focus on getting the best idea around on the table. Frequently it may not be yours. But you're also a key person in this decision, so you should make yourself useful with all the knowledge you have on the matter. So, instead of fighting for an idea, if you really care about outcome, what should you fight for?

    Fight for structure.

    1. The best idea is measured by outcome. So, first, define what success looks like. What will you be able to improve in the company if the winning idea is correct?
    2. Then, how will you measure this success or not?
    3. Then you can define rules of what is possible and not (budget, prioritization of resources)..
    4. Finally, trickle down into the operational details like implementation, review and tracking.

    After you have these few criteria written down, get approval from your manager or peers. The variables written on the structure (those 4 points) should somehow mimic how your manager decides himself (or should be deciding). Then you can populate some examples and send around.

    I've taken part at restructuring KPIs within an operational area of my company, the area responsible for deliveries. While everyone presented their KPI suggestions, it became bigger and bigger the number of KPIs to track, and it became also clear many of them weren't actionable. Number of deliveries made, per day and per person were two KPIs suggested. While those seem nice numbers to track, they are not proper KPIs for an operation, as they don't measure performance, they measure output. An operation can be awesome even if the number of deliveries falls on day, it depends among other things on how much marketing you did and sales you got. You cannot blame a dispatcher for doing less deliveries if his deliveries are harder, or if you give him a thinner pipeline. Etc. After this problem was pointed and assimilated, we set the objective (increase performance of Ops in Delivery, among others), the condition for success (KPIs being actionable, trackable and fair - in a word, performance-based) and only then did we start to break it down in variables, etc. The rest of the process became fairly autonomous and simple.

    When such structured discussions happen you have your team in synch. People will love you for including them and will value you showing your rationale. They will still argue but they will know you did a great job in helping them structure their own thoughts.

    This will save you time and headaches. It also helps your team understand and improve how decision making works, and they will trust your future opinions a of more. Because you don't fight for your ideas, you fight for structure.

    One cell to rule them all Sun, 21 Dec 2014 00:00:00 +0000 Whether it is the Microsoft or the Google flavor, spreadsheets are the beginning and end of a vast majority of business processes.

    They're used for many things

    • inventory tracking
    • building accounting tables like income statements and balance sheets
    • lead generation and tracking
    • creating business models
    • storing HR records
    • tracking hiring candidates
    • managing to-dos
    • planning marketing campaigns
    • building dashboards

    Why? Because they are very visual and easy to use

    • They use spacial relationships to define program relationships. We Humans have highly developed intuitions about 2D spaces, and tables come easily to us.
    • They are forgiving, allowing partial results and functions to work. One or more parts of a program can work correctly, even if other parts are unfinished or broken. This makes writing and debugging programs much easier, and faster.
    • They allow for secondary notations. The program can be annotated with colors, typefaces, lines, etc. to provide visual cues about the meaning of elements in the program.
    • They have many functions that perform transformations on data or groups of data, like finding the minimum on a list of values. They also have extensions that allow users to create new functions can provide the capabilities of a functional language.
    • They have collaboration in editing, and can be shared too.
    • They are ops-free, that is, there's no compile or deploy.

    All in all, they're especially great for 2 things:

    1. to build custom tools by yourself. Programming applications in standard languages is very hard.
    2. to build tools that change a lot. Maintaining applications is technically harder for a lay person. Even if you have a pro, organisationally it will be hard to support growing and changing requirements in the longer term.

    But if you abstract them, you can see how close cells already are to a program or a computer.

    A Spreadsheet CellA ProgramA computer
    Precedents/ dependentsTriggers/ side-effectsI/O
    Value (displayed)UII/O
    Value (stored)Database/ persistenceStorage

    They're currently a bit lousy on I/O and triggers. For example, there's no simple web requests and services, and little to no automation (machine interruptions).

    I am fascinated with the idea of using the structure (morphology) and function (physiology) of spreadsheets cells as the basic building blocks of programs. I'm convinced this would rapidly make millions of people happy. They would be empowered to really own their tools - to build, test, tweak, distribute, fix, share and improve them.

    We need Lego and Minecraft kind of flexibility. We want it. And so we will have it!

    The 3-wheeled bicycle Sun, 26 Oct 2014 00:00:00 +0000 Bikes are built for speed, just like a startup. Now and then I hear people going on crusades for adding features and business lines.

    From most angles, adding a 3rd wheel makes sense: more stability, novelty and differentiation, not extremely expensive.

    This has some practical business benefits: more people can use it, it's easy to market ("the bike which doesn't let you fall"), and you can increase price!

    But it it's not a bike most people want. It's not what they need. A bike is a great exercise of focus in providing fast, affordable, personal transportation.

    The temptation is always there to over-feature a startup's product, but there are problems which don't deserve a solution. It's critical to infer the product-market fit, but if you go around explicitly asking for customers what they want and to list problems with your product, you may end-up solving the wrong problems.

    I personally always refer to the vision. Does this feature move us closer to the vision or is it a nice to have? More frequently than not, the answer is No.