Question the Thesis

I originally wrote this for new team members at Entrepreneur First, but it applies to all startups. I’m going to share a guilty secret about how startups’ foundational beliefs come about, including the story of ours, and explore what it means for startups as they get older.



We’re going through a lot of growth as a team and as a company.

Around 45 people work at EF now. This may seem like a small number if you work in a big company, but it feels like a big number if you’re trying to start one.

While our team is still quite small, our company is quite old. EF is almost six, and six years is a long time in startups. For most of those six years, we barely grew: until 2016, our full time team was less than ten people.

This means we have history. Our history is a good thing if you’re joining now: we’ve spent a long time figuring things out. But history can be a bad thing for startups: time makes accidents look like strategies. And theses – the beliefs on which strategies are built – get harder to question over time.

For EF to succeed despite our history, new team mates, you have to question the thesis.

Here’s how theses are created, why they should change over time, and why you shouldn’t be afraid to change them:

Creating a thesis

In startups, theses are created by history, not strategy. The most important force in finding a thesis is survivorship bias.

When you join a growing company, you have to believe it’s right about something. If it’s not, it probably shouldn’t be growing. The more important a thesis looks, the more true it seems to be. But this isn’t a reliable inference, given how startups actually start.

In the beginning, you’re just trying to find something that works. You have a general hunch about what to work on, but you need something concrete. So you try stuff. Depending on what seems to work, you double down. What works becomes a thesis.

Take our focus on tech, something EF’s reputation is now built on. Here’s how it happened:

For EF, the general hunch was that starting companies would get more important for certain types of people. So we funded the best people we could get, by conventional wisdom: 30% technical; 30% sales; and 30% design people. The last 10% made up the numbers. Almost everyone was fresh out of undergrad because older people were hard to convince.

It turns out that technical people mainly wanted to work together and sometimes started promising companies when they did. Most of the sales and design people failed and got stuck if they were on their own. So, by EF3, we took only technical people. Soon, EF’s premise became ‘deep tech’.

Deep tech looks like a thesis now because it is fashionable. But really, computer scientists were just the people EF worked for when we had no idea what we were doing. You don’t need to be a computer scientist to start a company, you just needed to be a computer scientist at EF in 2013. We could have made a different conclusion – the programme is broken – but we doubled down. It was just a relief to find something that worked.

Changing a thesis

Once you find something that works, you build your reasoning and strategy around it. Despite this, you must constantly update the thesis you build on.

Our reasoning became that through technology you can be young and do amazing things, which is hard in other fields. We built a strategy: we are now extremely good at helping technical people build technical companies. We were right – young computer scientists can build extraordinary businesses, and we now have lots of evidence for that.

But you need to make sure your theses are always positive: young computer scientists can start great companies, but that doesn’t imply old designers can’t. Your early theses will only be a subset of what works.

In the beginning, a thesis is an accident of circumstance. And circumstances change.

One of Matt’s (EF’s CEO) most useful – although slightly annoying if you’re on the receiving end of a sudden resource – approaches is ‘resource based entrepreneurship’. Basically, you do different things depending on what resources you have access to.

At a superficial level this is obvious: to do more stuff you can either be more efficient or get more resources. But it’s not really about doing more, it’s about understanding which bits of your thesis are true because of your resources.

If we had had more experience, or money, or brand when EF was getting started, what we did would have been different. Our theses might have turned out differently. We might have been able to advise more experienced founders, change conventional venture funding, or recruit in larger numbers. Come to think of it, we can now do all of those things.

So why don’t we be more bold? What stops us questioning our theses? Once you know its history, a thesis become less precious. Once you have new resources, the thesis that got you them is up for grabs.

Without history, the aim isn’t to exploit the thesis you already have, it’s to find the right thesis to exploit.

Practical issues

Questioning the thesis is hard to do, especially if you’re new to the company. There are lots of practical issues.

The early team have to storify theses and history to get great people to join them. The story you tell is the one most likely to get people to join that’s retrospectively plausible. And survivorship bias doesn’t make a compelling story. So if you’ve just joined, you’ve probably bought a story created in the interest of getting you in and keeping you here. The historical team doesn’t always share everything.

This snowballs stories which sound good and support the theses, irrespective of truth. The faster you grow the faster the stories snowball. As the average amount of company history the team has directly experienced goes down, stories become more influential than what actually happened. After a while, it’s easy to forget the truth even if you were there.

The people most able to spread stories don’t get called out on them. A couple of months back, Matt was practicing for an important presentation and I gave him feedback on the content. He said he doesn’t get feedback like that often anymore. Why? Because his word is gospel: Matt has all the weight of history; whatever he says is assumed to have something behind it, even if it’s not obvious what that is.

It’s hard to question the thesis without trust. Some trust takes time: the trust I have with Matt by now is part boss, part friend. But you don’t need to be friends with Matt to critique him or his theses. Like everyone else, he wants to do the best job he can and, of all people, he is most incentivised to find out if his theses are wrong. You don’t have to trust Matt to trust his incentives.

Unfortunately, as you get more responsibility people find it harder to question you. People don’t want to question your theses, and don’t feel your development is their responsibility, even if both affect their outcome. We expect our leaders to improve us, but not to have to improve them. As a leader, this is terrifying. More power with less feedback is a dangerous thing.

Over time bad theses will get harder to remove. There is a bias towards replicating the behaviour and beliefs of the early team: people infer that the people who’ve been around the longest have survived for a reason. In some ways this is true, but in others they’re just getting away with it.

Your job is to find out which. And the only way to find out is to question the thesis.

Thanks to the EF team for reading drafts of this. If you’d like to question our theses, start here.