Will Accelerators Accelerate or Decelerate Startups?

In Dotcom 1.0 there were Incubators. These businesses typically rented space, facilities, admin and backoffice support to startups for some cash and a share in the success (or failure) of the business, typically back to backing it against investment from others who hoped to get into startup portfolio investment. Cometh the Dot Bomb, goeth the Incubators, and they got a bit of a bad name, a recent study by Jared Konczal of the Kaufmann Foundation, showing what everyone suspected empirically - Forbes:

The findings reveal that the effects of incubation are potentially deleterious to the long-term survival and performance of new ventures. Incubated firms outperform their peers in terms of employment and sales growth but fail sooner. These are important findings for policymakers who support incubation as a strategy to increase employment locally and for entrepreneurs who risk their livelihoods in order to earn a decent living.

Claims that incubators are highly successful and serve a significant number of businesses are overstated. The comprehensive process used in this study to identify the largest possible sample of incubated firms uncovered a fraction of the number of incubated ventures that supporters of incubation claim exist. While improvements are likely possible to the methods used in this study, this study roundly refutes the poorly documented and unpublished studies that cite much larger numbers of incubated firms and much higher levels of performance.

Cometh Bubble 2.0, cometh Incubator 2.0 - only they have been rebranded, "Combinator" was tried but never took off, the New New Word is an Accelerator. The business models has also changed a bit- Forbes again:

Accelerators are organizations that provide cohorts of selected nascent ventures seed-investment, usually in exchange for equity, and limited-duration educational programming, including extensive mentorship and structured educational components. These programs typically culminate in “demo days” where the ventures make pitches to an audience of qualified investors.

In other words they are Incubators with a more direct funding link (and often without the benefit of any facilities to help startup companies with). But are they any more effective?

Claims are made by their proponents that they are more effective (eg Grasshopper, here). As Konczal notes, the major criticism of such assertions is there is no "blind test" vs other startups that did not use Accelerators. In essence though, from the data, the main issue that emerges is that you have a huge differences in impact between the Very Early In to any trend (like Y Combinator) that work out very well, and the following horde of me-toos that don't (as was true for Incubators, and the VC game in general). One piece of research is noted by Konczal. Based on the c 120 Acceleraror programs in the seed-db.com database, it shows that by 2012 Accelerators are a massive wealth reduction agents to the end-of-the-line funders. As the chart at the top shows, splitting Y combinator out from The Rest, Y Combinator is profitable whereas The Rest, on aggregate, are massively loss making - $300,000 lost for every startup job created is all you need to know. (The author points out his data is not accurate, but has some sound reasons to suppose it is indicative - ie the non reported costs are probably larger owing to many types of investments being below the radar).

Interestingly, Konczal notes a NESTA (UK) study from last year that I read just after we were developng the Bubble-O-Meter, the NESTA research noted the rapid growth (see graph below), said that early evidence was that they were a positive influence, but noted 4 main risks:

(i) Good companies still fail after accelerator programmes (as Konczal notes we also don't know what the success rate of not going to an accelerator is)

(ii) They exploit startup founders (The amount of equity taken by accelerator programmes has also been controversial - ie they take a lot)

(iii) They attract companies that are already struggling (and this is what usually leads to the effects described above - Good companies still fail after accelerator programmes, as the number of accelerator programmes rises, they will struggle to avoid making investments in B-grade companies.

(iv) They’re helping to create a bubble (If accelerators continue to grow and start producing thousands of small companies, we can expect to see a bottleneck developing and in the event of a crash in confidence in the sector (as happened in the Autumn of 2008)

We've been following the Accelerator trend since 2007, here was what the state of our research c 2008 was:

I would make a few predictions here on the Next Generation of Incubators 2.0 (I'll call them hothouses as its more part of the zeitgeist):

(i) There will be little difference in average outcome between hothoused startups and organic ones, but later hothouses will underperform on average, as they will be lower quality.

(ii) There will be a higher propensity for the hothouses to follow fashion, as following the crowd is an easier sell to funders. This will make them more costly than organic plays which will by and large be earlier in.

(iii) There will be little advantage gained from using youth and inexperience early up, apart from low cost of the initial "throw at wall" phase - if anything sticks, age and guile will be required, and as it has less stake, it will be at premium pricing - or the founders have to give up a large slug of equity.

(iv) The "Next Web" will not be built in the hothouses, as innovation is not really their business model - fast following is.

By 2011, when NESTA was writing their report, we had seen the latest bubble in Accelerators and listed it as No 4 in our Bubble-O-Meter trends.

Looking at the huge growth in Accelerators since c 2006 again now, and juxtaposing the Kauffman data, we'd propose that bubble-like behaviour has indeed emerged here, and the me-too startup count is rising (empirical observation sure, but a lot of them do look similar). If the dotcom era is anything to go by, this implies that anything founded since the "bubbly bit" began in the graph - about 2007/8 looking at the graph, soon after Y Combinator according to the chart - is probably not going to make any money for its funders.