“What do accelerators do and are they effective” is a tough question to answer. The best ones are continuously (and quietly) iterating on their model as the funding environment changes along with their cohorts. When you try to compare the success of cohorts from different years you could be comparing companies that have a very different set of ‘tools’ applied to their ‘acceleration’ process.
In an article in HBR entitled What Startup Accelerators Really Do the author points to some research in attempt to explain what accelerators do and evaluate if it is effective. The evaluation is interesting.
..accelerators can have a positive effect on the performance of the startups they work with, even compared with other key early-stage investors. But this finding is not universal among all accelerators and so far has been isolated to leading programs. the author gets a little more specific than the HBR article on which programs make a difference that can be measured.
In the research paper the author gets a little more specific than the HBR article on which programs make a difference that can be measured.
We find that TechStars and Y Combinator are strongly associated with ventures being faster to raise capital relative to the matched comparison ventures. Effects for some accelerators (500 Startups, AngelPad, Dreamit, and Excelerate Labs) are lower in magnitude, while still other accelerators are associated with ventures that were slower (LaunchBox Digital) or no faster in their time to raise to raise (Seedcamp). In sum, these results suggest that some accelerators accelerate venture gestation, that some accelerators accelerate more than others, and that acceleration is difficult and not achieved by all accelerators.
What the paper suggests is that accelerators work but only two (YC and Techstars) are measurably accelerating. Other programs are not. The group of accelerators they looked at where limited as they looked for accelerators that had at least four companies with funding events before summer 2013 that are public in Crunchable.
Referring to the research paper the article does a great job of defining an accelerator and makes the point that there appears to be a secondary benefit of accelerator programs – they increase the overall investor activity in the community around the accelerator and not just for companies in the program.
Figuring out what accelerators actually do is hard
There are some challenges not discussed in this article that I would have liked to have seen mentioned: a) companies go into multiple programs over a short period of time, b) what is ‘normal’ for the average company, and c) picking the best of all companies that are applying to accelerator programs is different than acceleration as its defined.
a) Companies go into multiple programs.
Companies enter multiple accelerators over a number of years finding their way towards TechStars, 500 Startups, and/or Y-Combinator while maybe changing their name and/or product as they go.
b) What is normal?
There is no ‘control group’ — it is unknown what a ‘normal’ number of companies being founded is and whether they would eventually migrate to the valley or a larger centre with a larger network or if they stay put how would they ‘normally’ do? This opens up some questions on whether the presence of so many accelerator programs is driving the growth in investor activity or if it a result of the growth in investor activity.
Do early stage investors drive the creation of these programs as a way to manage their deal flow? Are there any companies left that choose not to go into an accelerator or at least apply to one? My guess for both those questions is that yes they are essentially managing deal flow for early stage investors and because of that there are few to no companies founded by first time founders that are not in an acceleration program.
c) Picking the best of all companies isn’t ‘acceleration.’
No one really knows what ‘stage’ a company actually is when it enters a program. The top programs could just be better at attracting the limited number of companies that are ready (but don’t know it at the time) to raise capital. This is related to my first point.
How do we know what works and what doesn’t?
In order to learn, iterate, and grow the number of successful companies there needs to be an understanding that building a company is an art not a science. There is no ‘process’ that will be repeatable by any two individual founding teams. The real ‘magic’ is found in the network — the peer group of masters and apprentices.
Probably the biggest challenge with accelerator programs is that they try to differentiate themselves in order to attract both funding and applicants. From specializing on a particular subject (sales, hardware, etc) to how they manage a cohort or funding.
Founders of anything benefit from building a peer group across all verticals and they should be co-located as much as possible with any/all barriers removed. No teams in different buildings. No walls. No classification and specialization. Just one big ‘cohort’ of people with a company that does anything.
Accelerator programs need to focus on grooming the larger ‘network’ and focus on the quality of those network connections. If they do it correctly they will find the startup world is small.
Y-Combinator gets it. They are years ahead of pretty much all accelerator programs having made a shift away from what people often referred to as ICT to focus on everything and anything that they have a partner interested in. YC’s power is in its network of Black Swans.
What could this means for programs today? It is very likely they are not as effective as they think they are at accelerating companies. Don’t rest on positive media mentions and good marketing. Programs should avoid following a model thinking there is a ‘best practice’ to follow and focus more on the success of the companies they are helping.