All your predictions are wrong

Don't get stuck on your beliefs

A non Socratic intern

As an entrepreneur, you may come from diverse backgrounds. It's likely that you worked in a relatively large company before, where managers asked their associates to create predictions for six, twelve, and thirty-six months and develop complex, well-designed business plans. If you are the honorary possessor of an MBA, you may even make a living from this. Sorry to burst your dream, but this is all bullshit.

Hey, my dear associate—or free intern—could you tell me who our ideal client is? What are the corresponding market shares? How much revenue will we generate using the sales strategy that your other free colleague will create? The story from there is simple: build a model using open-source data or sales data from another department. Add a few explanations if things don't go as planned. Convince yourself that your model was correct simply because you sold as much as you projected. A million possible models could have led to the same result. That’s corporate life, and if you are reading this article, you probably don’t like it very much.

I’m the boss now

Now you own your start-up, you are the CEO, and god damn it, you will show the world your product is ready for selling. You have to do it yourself, with no free intern and no six months deadline to deliver a hundred and fifty pages power point and just go by the -boring as hell- book. You shall do it now if you don't want to make your bank account leak like a Russian submarine¹.

For you, my fellow entrepreneur, I have a piece of bad news and a piece of good news that you may not be ready for. The bad news is that everything you know about your potential clients before talking to them or selling to them is likely wrong or useless. The good news is that, in this case, it's actually beneficial to be an ignorant bastard. Yes, it’s excellent news, as long as you love science. You should believe in short, repeated feedback loops and hypothesis testing rather than sophisticated predictions.

Before we dig into the systematic, iterative method to build your company, let's first understand why almost all your predictions are relatively useless. We’ll explore why what seems obvious or logical often is not. Let me share a personal story.

Shiny features get you nothing

One of my main projects was, and is still to this date, an accounting platform offering accountants and bookkeepers the ability to collaborate with their clients and automate redundant tasks. Suppose you come from a well-developed and very digitalized country. In that case, it might sound very dull. Still, in most emerging markets, people do their accounting closure on paper and barely use Excel. In the common case, they are using software, it is this forty-year-old black and white Windows 95-style desktop program or some shady cloud applications that explode every time a developer pushes an update.

They tend to call those digital nightmares ERP. Here I come, heroically changing the fascinating world of bookkeeping and selling this fancy feature: why not stop entering everything manually and use OCR (optical character recognition) technology? How could accountants say no to gain some time? How could business owners not be happy stop writing every single items of their invoices so they can have an inventory and a treasury statement up-to-date?

My predictions were robust, and I managed to sell my software to a solid group of accountants, convinced that my algorithm was the critical factor in these successful negotiations. Obviously, accountants were looking forward to entering the digital age and stopping being considered corny suckers. How could that not be the case?

A few weeks later, after managing to get a few users using some strategies, we will see later - it was still a relatively early stage project - I decided to ask my clients on their feedback. It was pretty wise, as I am a technical founder, to check on the software usage by my new clients.

They must be processing invoices and start thinking about how their life could have been had they had this OCR a few years before. Maybe they would be fitter, smarter, and their girlfriends and boyfriends more attractive. Do you know, after a month, how many invoices had been processed by my dozens of accountants and bookkeepers?

0

Yes, zero.

Yes, that's the number. None of my clients had actually used the algorithm. In contrast, it was the flagship feature in my speech and, up to this moment, the main reason they were changing their old software for my shiny application. The good news is that they were actually quite satisfied with my product, but it was for entirely different reasons than I had thought in the first place. Even a boring-looking profession such as accountant can give you some surprises.

Boring stuff gets you everything

As you can imagine, I was pretty surprised and decided to ask my clients what features they liked the most in my software and what made them happy. As you may know, accountants have to generate some monthly reports and send them to the tax agency or the business owner, some related to all sorts of taxes or specific calculations regarding some accounting units of the business, funny stuff here.

It turns out that some extraordinarily boring report has to be done almost monthly. In most software, the accountant must visit the reporting section, choose from an endless list of boring reports, and pick the one to generate. In our case, the most frequent one was on top of the welcome page, with a star, and … well, that's it. My clients were thrilled with how the reports looked like, and they could generate them in two clicks instead of six, which is a marginal gain of time on a monthly, if not quarterly, task.

I missed the point by thinking my potential customers were looking for a technological revolution: they were not. They were only looking for something that is not awful and has a decent user experience. The innovative part would come way later. After a lot of iteration on the way to sell and deploy it, they use the shiny features that make them gain their time back. As a founder, I had some biases I fiercely injected into my market strategy, and you are probably doing the same.

Don't think you know your customers before you sell them something.

Don't think your customers know what they like about your solution before they use it.

Don't make predictions. Just get some data and test some hypotheses.

Keep the faith,

Voss

1. If you are VC-backed, you don’t need it, that’s why you don’t know the real struggle of entrepreneurs. Sorry, spoiled kid. (I’m raising funds btw)