Interview with Matt Clifford


In 2011 Matt Clifford and Alice Bentick left their consulting jobs in McKinsey to found their own nonprofit, Entrepreneur First. Their idea is to take about 30 talented and ambitious graduates each year and help them form teams and start companies, in order to promote entrepreneurship as a career path. They have already achieved some impressive results with their first cohort group of 32 graduates founding 11 companies which collectively achieved a market valuation of £22 million in their first funding round.

As part of our research on the best routes into entrepreneurship we talked to Matt Clifford about his thoughts on the question.


Key points Matt made in the interview:

  • A high level of technical skills seems to be the most important single attainment of someone interested in becoming a tech entrepreneur.

  • Startups inevitably involve failures and things going wrong, so determination is an important trait for entrepreneurs.

  • If you are interested in starting a startup then you should consider developing domain expertise in a sector other than tech that is ripe for disruption.

  • You probably shouldn’t do an MBA before before going into startups unless you would do one anyway.

  • While many people leave consulting jobs to start startups, they do not provide ideal preparation for most startups.

  • It remains hard to find good data on indicators of success in entrepreneurship, current research is more qualitative and indicative.

Excerpts from the interview:

1) On your website you say you are looking for graduates with skills, track record, ambition and determination. We’re interested in what kinds of more concrete indicators you’d look for to judge these?

The only way we’ve found to do this fairly successfully is to ask competency style questions, found in a conventional graduate program. “Tell me about a time when …” We phrase those so that it gives people an opportunity to interpret them quite broadly. That’s very important for us because one startup is very different from another, so something that is very relevant experience for one might not be very relevant for another.

The way we try to introduce a bit of rigour into that is that each person is interviewed by at least four people, some of whom will repeat questions. Calibration is something that we think about a lot and one of the ways we deal with that is that we get people to interview separately, mark separately and assess each criteria separately. Otherwise it is very easy to be drawn by charisma rather than what you are trying to assess for.

2) Could you rank in terms of preference for acceptance into your program: learning to code, academic learning (doing well in class), working in a startup, summer internships with big companies, attempting a startup, getting a sales job.

I suppose the honest answer is it depends on how extreme their attainment on each of these dimensions is. At a median level or a standard deviation above the mean, the most important for us would be technical skills. The rest of it is much more teachable in a short period of time. Learning to code is hard to simulate and hard to teach in the short time we have available. This very much applies to our niche, which is software/tech startups. They’re very product focused. It’s not just because great products are necessary, but because speed of iteration, speed of learning and turning learning into product insight seems to be what makes for good companies. If you’re having to outsource all your coding, you might learn something new from a customer every day but you probably can’t afford to iterate your product more than once a month or once a week. Our best teams reiterate their product daily which is almost impossible if you don’t have the tech capabilities in house.

Attempting a startup in the past is probably the next strongest. Firstly, it shows intent. It is easy to take really smart people who look really great on paper but it turns out they don’t actually want to start a startup. The Social Network movie had a huge impact, way more than you would think possible. Lots of people think they want to start a startup but have no idea what that it means. It’s very unglamourous compared to what people think. Secondly, obviously, they also just learn a lot from that, particularly about what goes wrong and how to avoid it.

If they failed in the past that would be totally neutral above and beyond the fact that they tried. What we would really want to know is what they actually did. Failing because you couldn’t find a fit between the market you were trying and the product you were creating: that’s fine. That happens all the time. Failing because you haven’t really tried may be an indication of a more flaky personality which is a real problem.

Working in startup is also really good. It also covers the intent and the learning. It can also mean you are also bringing a network with you that will help you succeed.

We’re not massively concerned about academic learning. We have a joke that we either want people with a 1st or a 3rd, but we don’t have enough data to back that up! We obviously like really smart people, they tend to be good at problem solving, but not everyone spends their time at university optimising for their degree, and that doesn’t mean they won’t be good entrepreneurs. On the other hand if you spend your time at university optimising for your degree we want to see that you’re good at that. Partly it’s just a question of goal achievement. But so far, it doesn’t feel like there’s a correlation between academic success and outcomes.

We’re really not bothered about summer internships in big companies either. The only benefit is that with competitive programs getting into them shows that you are driven and smart. Even then, I would then want to believe that they really know what they are getting themselves in for if they think they want to start a startup, because by themselves driven and smart are not enough.

Getting a sales job can be good. Some sales jobs can be amazing training for being an entrepreneur, because startups are about selling. But if you’re used to being only able to sell if you have unlimited resources then startups can be a bit of rude shock, so it depends on what you’re selling, what channels you’re using, etc.

Finally one thing that isn’t in your list that we rate extremely highly is evidence of determination. We find that whether it is picking a team, picking an idea, choosing a product, choosing a sales channel or choosing a market strategy, the first thing you do will fail. So whether that completely rocks you or doesn’t bother you in the least makes a huge difference. Also, just looking at the people we’ve worked with, the biggest cause of failure is people giving up rather than running out of money. Some of our biggest successes in fundraising have been with companies that we’d almost given up on, though this is more anecdotal.

One thing we do to test this is a fairly abstract exercise which is based on real situation that we faced:

Someone had given us a free office and we had two weeks to fill it with 30 desks and 30 chairs and nearly zero budget. We tell this situation to people and ask them what they would do. Then, whatever they tell us we say “Cool, that got you 10 what are you going to do to get the next 20.” Then for whatever they say next we say “that got you two”, “that got you two”. We only did this last year so I don’t have evidence that this works yet but it feels like it makes a difference. It would be hard to fail on this answer just because your ability to solve that particular instance is not going to make or break a startup. However a very good answer on that can put you in the top tier where you weren’t before. It shows a side of your personality that might not be obvious. Some people give extraordinary answers that you just can’t guess.

3) Why do you think these traits are important? Do you have any data from your program so far or other studies that back up your choices?

We have very low-level micro data on our last cohort and their outcomes. There’s also academic studies on successful entrepreneurs and how they differ from the population. When we first did this we did a lot of searching on JStor and Google for this. I would say the evidence is still indicative and qualitative.

Entrepreneurship means so many different things to different people and the niche into which we are recruiting is so niche that I think it requires a qualitative interpretation of previous research. Why do we look for tech backgrounds? It’s not because there’s a whole lot of research that people with tech backgrounds are important. It’s just that we are going to get people to build tech products so if they can’t build tech products they can’t do it!

More generally it is hard to find good data in this industry. When you look at the data from various programs there is massive unobserved variable bias. You’re selecting on the dependent variable and you never get to see what the outcomes of the people you didn’t select would be.

We try to do a lot of looking at what indicators early on the program predict success, but we as an organisation have a very slow iteration cycle. It will be hard for us to ever run more than two programs a year, and hard for that to reach more than 100 people a year.

YCombinator have got a lot of data they’ve collected and they’ve done 600 companies. But they’re quite precious about it. They see it as a commercial secret. And some of the stuff they have put out public has proved to be very controversial.

I’m very strongly committed to the data intensive approach. But no one’s met the bar I’d be happy with on that side.

4) For people who want to become entrepreneurs without going through EF, what other steps do you think are particularly useful? For example, is it useful to do an MBA? Is working in VC (venture capital) a good step? Or do you think it’s better just to network with people in the sector and start working on something?

I wouldn’t do an MBA unless you would do one anyway. In fact within the community it seems to be quite a negative signal. I think the reason is probably that MBAs teach you how to run large businesses and startups are not small versions of large businesses. They’re wildly different so most of what you learn is not that helpful.

Working in VC can be helpful but interestingly one trend in the last 5 years is the opposite. Instead of VCs being a breeding ground for entrepreneurs it’s more that entrepreneurship is a breeding ground for VCs. It’s relatively hard to get a standard role in VC now. It used to be that you’d spend two years in banks or consulting, do your MBA and then going into VC, but most VCs are avoiding that model nowadays in favour of hiring successful entrepreneurs, or even unsuccessful entrepreneurs.

If you want to be an entrepreneur, you don’t have an idea and you don’t get into a program like EF, I think, go find a problem that needs solving. I think you usually do that by going and working in industry. I would develop as strong a domain expertise as possible in the sector you want to disrupt. Not the tech sector. I’m assuming that nearly all successful startups now are going to be in some existing industry overlayed with tech. And I think unless you have some domain expertise or you have a network that can help you accelerate your acquisition of that domain expertise, which is what I think Entrepreneur First provides, then it strikes me that the best way to do it is not to be a banker, not to be a consultant or one of these really generic jobs, but actually go and work in the industry you think is ripe for disruption. Keep an eye on the tech side, keep an eye on the entrepreneurial side, but when we look at non-technical founders, they’re much more useful if they actually know what the real problems are in their industry rather than if they’re great sales people.

To be successful you have to build something that solves something someone is already spending their time thinking is a problem. It’s hard to convince someone that something is a problem, they kind of already need to believe that they have one. So if you’ve been one of those people it is a whole lot easier.

I think what EF is really good at is bombarding you with these people so that you actually know what to build. One of our most successful companies from last year, Kivo, essentially build powerpoint software for management consultants. They’ve never been management consultants but we know loads of management consultants so they just spent loads of time with the people we know and got to the same stage. But I think if you’re doing it without a program, like EF or like an accelerator, than I do think you probably want to have that knowledge yourself.

The tech route makes sense if you’re heavily technical, you have an idea, and you can just start building something that people do want. Something like Dropbox, which has been very successful in the last 4 years, or Facebook, fits in this model. Some consumer product that the founder wanted themselves. But if you get into enterprise software, that’s where domain expertise is much more useful. An example is Salesforce which is a huge company which came from a guy who knew sales who said “Wow, we need better tools for this!” and we’ve seen this in edtech, where it’s teachers, healthtech, where it’s doctors essentially saying we need better tools for what we do. We’ve also seen this in a range of other sectors from mining to manufacturing to distribution and logistics.

Alternatively work for a startup with someone who is prepared to spend some time mentoring you. That’s the opposite way around, but if you do that you will have the other side of the skill set built out by the time you start your own thing. There are a lot of examples of Google product managers leaving to start startups. The people who were involved in Paypal early on have created dozens of new businesses between them that have gone on to be very successful.

5) How useful was your time at McKinsey for starting your own organisation?

For me it was very useful. It would be relatively difficult to run EF without corporate experience. Even now we’re not building a tech product. We’re building kind of a people product, and a lot of what we learned at McKinsey has been helpful for that. But on the other hand we also had to unlearn a lot of things we had learned. Within certain sections of the community it probably gave us less credibility. In general it’s not great prep for starting a startup. That’s one of the reasons we built EF. So many people who go to McKinsey leave to start startups, (about a third of the people in my year did it) but if you were designing something to prepare people for starting startups it wouldn’t be McKinsey so what would it be? I hope the answer is EF. Why do so many ambitious people who want to start startups go to Mckinsey? It’s hard to answer, but I hope part of the answer is because EF didn’t exist.

6) How easy is it for international students to get permission to work in the UK for your program?

Visas are something we struggle with right now. At the moment we can only take people if they have the right to work in the UK independently. We hope for this to change in the future.

7) How important do you think location of a startup is for success? e.g. How does Tech City compare with silicon valley for tech startups?

Location very important for network. Density of networks is absolutely vital. You need to be able to find people who will mentor you, invest in you and can help you out. It’s a lot better than it was 2 years ago in London, it’s a long way from Silicon Valley, but it’s certainly at the stage that it’s totally viable in London.

8) What are the best other materials you about how to become an entrepreneur and who is most likely to succeed that we should check out?

There’s some very practical advice that Paul Graham gives in his essays which are available on his web site, which and are very widely read in the industry. The new thing that has become very popular and been very successful has been The Lean Startup by Eric Reis. I think taking these two as a starting point is the best bet. Also we’ve found Noam Wasserman’s The Founder’s Dilemma useful in designing our selection process. It’s quite academic, but packed full of great data!

If you are interested in finding out more about Entrepreneur First or applying for their program be sure to check out their website.