Five baseline assumptions on seed accelerators

There has been an explosion of new seed accelerators recently, and with that comes an explosion of press interest, blog articles, and more. I was even interviewed for a recent article by the Wall Street Journal, though my quotes were all cut. (Speaking of which, I learned a lot of about media through that experience. Sometimes it doesn’t matter what the writer/journalist thinks, the editor really calls the shots and enforces a point of view.)

I want to lay out the five baseline assumptions that I make when explaining seed accelerators:

  • All accelerators improve the chances of startup’s success
  • There is a wide spectrum to how much help an accelerator gives a startup due to reputation, deal flow, quantity/quality of applicants
  • The best seed accelerators are started in order to do better investing, and partners really make money through follow-on investments
  • A lot of seed accelerators will fail, and there’s no problem with that
  • Entrepreneurs should understand what they’re selling to an accelerator (but not worry about it too much)

And here’s why:

All accelerators improve the chances of startup’s success

While I don’t think that every startup should go through an accelerator, I do believe that going through any accelerator improves the probability that that a startup will be successful. It’s very straightforward: the whole point of the accelerator is to provide tools, mentorship, connections, and support to startups so they can be successful. Perhaps I’m a bit of an optimist, but I would be surprised if any startup coming out of an accelerator would say that the program had hurt their chances for success or provided them with zero value.

Unfortunately, this is a very difficult data point to test. I’ve assembled a lot of information about the status and success of over 1100 startups that have been through nearly 100 accelerators across the globe, but capturing the data for the entire startup ecosystem that hasn’t gone through an accelerator is damn near impossible.

There is a wide spectrum to how much help an accelerator gives a startup due to reputation, deal flow, quantity/quality of applicants

I doubt this is a controversial point. The degree of acceleration that startups get by going through Y Combinator is significantly different than many of the newest accelerators. For example, YC companies have a very high rate of getting follow-on funding, and their alumni network is unparalleled. Newer programs don’t have anything like this.

When it comes to understanding the seed accelerator ecosystem, the best metaphor is the US university educational system. There are some programs that are clear leaders; like Y Combinator / TechStars and the Ivy League. There’s massive demand to get into these handful of programs, and they attract the very best startups. Once inside, you have some of the best access to people in industry and massive advantages because of that credential. And because they’re so good, they attract the best people with the best chance of long-term success.

There are also more regional accelerators, like the state university system. They tend to be more attractive for people looking to stay closer to home. Some of these have the capacity to be real powerhouses, but many are just average. But even though I call them “average”, by graduating from the university/accelerator, you have a better chance of long-term success than if you did it all on your own.

Finally, I like seeing the growth of vertical-specific accelerators like RockHealth and Imagine K12, which are like professional schools. They provide real benefits for companies in verticals that have very specific needs and very specific markets. It’s no wonder you see companies like Agile Diagnosis go through Y Combinator and then later on also decide to go through RockHealth. As I mentioned in my thesis three years ago, I believe these specialized accelerators are where some really great and successful programs can and will be built.

The best seed accelerators are started in order to do better investing, and partners really make money through follow-on investments

When I interviewed Paul Graham and David Cohen for my thesis back in 2009, one thing was very clear to me. They started Y Combinator and TechStars in order to do angel investing better. In short, Y Combinator and TechStars were primarily founded to make money. This aligns the program and the startups; the startups want to create successful businesses, and the program partners want to make money which can only happen if those businesses are successful. Because they’re judging and assisting startups solely on their product/market fit and potential, I believe they have the greatest chance of generating the exits required to become profitable.

I’m personally not a fan of accelerators that are started for the purpose of, say, “kickstarting the startup ecosystem in Northeast Montana.” They usually get some government funding or grants from organizations that they’ve convinced to fund them. In short, these programs are often funded for ego. Now I don’t think these programs should die, because of the very first point above. (Any program will improve a startups’ chance of success.) But the incentives are not necessarily aligned in this situation. Startups still want to create successful businesses, but accelerators want to prove they’re doing good things for the community. I believe that these programs will not typically generate startups with exits that will make the program profitable long-term, so the program will need to go back to their investors to continue.

No accelerators that I know of do any follow-on funding, though some programs are able to offer convertible loan notes of small size. This limits the profits that the accelerator will eventually earn. But this hides the fact that the partners of the programs DO invest in follow-on rounds individually through their own personal angel investing. For example, I understand that many/most of the YC partners invest in YC companies, though from what I’ve been told they don’t lead any follow-on rounds. (Probably partly for signaling effects, and partly because they have a lot of demands on their time.) So while the accelerators’ stake is initially 5–10%, it’s diminished with each round of funding. The partners who invest personally are able to continue to invest in future rounds, and while I have yet to do an in-depth analysis here, I believe that’s where the partners REALLY make their money.

Leading on from that, what an accelerator really brings to program partners is deal flow. So even if a program is marginally profitable or unprofitable, the founders/partners could effectively use it as a loss-leader for much more profitable follow-on investing.

A lot of seed accelerators will fail, and there’s no problem with that

I’m modifying this phrase from Pascal Finette’s blog post. A number of accelerators are started for the wrong reasons (ie, not strictly to make money) so I believe that they do not have a bright long-term future. I agree with Pascal that the money many programs earn from exits will never be able to make up for the costs of the program, and many will eventually close.

Seed accelerators failing isn’t bad! Even if an accelerator that’s trying to kick-start the startup ecosystem in north-east Montana runs for two years and then dies, that’s two years of startups that have been trained and two years of connections between startups and investors and mentors. Sure, while it could have potentially had a better outcome, that’s not a bad result. From my data, just under 10% of the seed accelerators I’m tracking have already failed. (Though some have been resurrected in other forms.)

Who really loses when an accelerator dies? It’s not the startups; they’ve already gone through and had their experience. The only people that lose are the investors behind the program, and they should have understood what they were getting into before they began.

Entrepreneurs should understand what they’re selling to an accelerator (but not worry about it too much)

When I was speaking to the journalist from the Wall Street Journal, she mentioned her editor was very concerned about the cost of these programs, that startups are giving away too much equity. I believe this line of thinking is short-sighted and wanted to briefly explain why.

The default end-state for a startup is failure. The reason for an accelerator’s existence is to help prevent failure, and they take a small equity stake in companies in order to align their incentives. The accelerator only ever “wins” when their startups “win”. Paul Graham has (of course) written on this topic, and made an easy mathematical argument. You should be willing to give up X% equity if afterward the business has a (100/(100-X))% better chance of success. So if you sell 10% of your equity to an accelerator, you’re better off provided that your business is (100/90 = 1.11) 11% better off than before you went through the program. I think with the vast majority of matches between startups and accelerators this price is most definitely worth it.

There is a market “price” / valuation for different programs, and entrepreneurs should recognize what they’re selling. For experienced entrepreneurs to go through a new accelerator is perhaps not worth the cost, because it could duplicate their existing experience, adding no real value. That’s why these programs should (and do) specialize in new entrepreneurs. Top-tier programs are able to add value even for experienced entrepreneurs, so it’s not uncommon for founders to go through Y Combinator even after starting and successfully exiting a previous YC company.


Seed accelerators are great opportunities. For beginning entrepreneurs virtually any program would improve their startup’s chances for success. For beginning program founders, I would suggest that you deeply examine your motivations, and what could/would happen if the program doesn’t get the exits you expect from the first 1–3 years of startups.

There will be more programs founded, and a lot more startups going through them in the next few years. I hope the information here provides a good framework for discussing seed accelerators in the future.

%d bloggers like this: