Innovation Accelerators: Defining Characteristics Among Startup Assistance Organization

C. Scott Dempwolf, Jennifer Auer, and Michelle D’Ippolito, October 2014


While accelerator programs have been around for almost a decade, their more recent successes have catalyzed a surge in their popularity among investors and entrepreneurs. Given the potential for these programs to grow scalable high-value startups quickly, different groups of policymakers are now starting to consider how to apply the accelerator model to meet public policy goals. However, accelerator programs are not guaranteed successes.

Many questions surround accelerators’ application to broader arenas, and there is a shortage of robust sources of data or metrics to evaluate their efficiency and effectiveness. This report aims to help entrepreneurs and policymakers to start answering these questions by categorizing a variety of startup assistance programs to determine what factors distinguish accelerators from other programs. Using these distinguishing characteristics, the report provides a robust definition of accelerators as well as a starting point for developing meaningful metrics to determine the relevance of accelerators for policymakers.


Building on recent work by Cohen and Hochberg, this report defines accelerators as:

Business entities that make seed-stage investments in promising companies in exchange for equity as part of a fixed-term, cohort-based program, including mentorship and educational components, that culminates in a public pitch event or demo day.

Accelerators are an innovative startup funding mechanism leveraged heavily in the tech sector. These programs use a selective application process to target scalable, high-value, and high-growth startups. Accelerators help entrepreneurs commercialize sometimes underdeveloped business ideas, helping startups to go public, get acquired, or receive additional funding in a brief span of time. There is evidence that in some cases startups that graduate from an accelerator get funding faster than those using alternative funding mechanisms. The accelerator does this through educational and mentorship programs which can be extremely useful to entrepreneurs. More significantly, accelerators can connect startups with networks of other entrepreneurs and potential investors giving program participants invaluable social capital contributions. These may be the biggest benefits of participating in an accelerator program, since they connect entrepreneurs with a sizeable pool of potential investors. Recently, investors have started to embrace the accelerator model as a way to distribute the inherent riskiness of investing in tech startups over a large startup pool.

Successful accelerator programs may not be representative of the model in general.

Y Combinator has been one of the most successful accelerators to execute this model. It has graduated multiple startups that have not only changed seemingly established industries but also in some cases received billion-dollar valuations. Notable startups that have participated in the Y Combinator accelerator include Reddit, Dropbox, and Airbnb. In response to these highly visible successes, accelerators, largely in the tech sector, have sprung up all over the world.

The non-profit and public sectors have started to take notice of the success of accelerator participants and are hoping to utilize the accelerator model to work toward meeting public policy goals. This model is particularly attractive to non-profit and socially responsible startups who may find it difficult or inappropriate to receive funding from venture capitalists. Two examples of innovative “social accelerator” programs are the ARK Challenge in Northwest Arkansas and Conscious Ventures Labs, a benefit corporation located in Howard County, Maryland. Both of these accelerators support groups of non- and for-profit companies that have clear public welfare goals as part of their missions.

However, despite the prominence of a handful of success stories in the tech press, these exemplars are often the exception and not the rule. The idea of using accelerators as a policy tool to grow the next Facebook and jumpstart a local economy therefore relies on very low-probability events. Due to their abnormality, venture capitalist Aileen Lee lightheartedly calls startups with billion-dollar valuations “unicorns.” According to her calculations, as of November 2013 only one in every 1,538 startups (about .07 percent) founded within the last decade can be classified as a unicorn, and none were founded in the last couple of years. Moreover, it took “unicorns” on average seven years to get to a liquidity event (filing an IPO or being acquired). So even if a startup is able to grow quickly, it may take many years to realize financial value from that growth.

Therefore, while accelerators have certainly been involved with startups that have become billion-dollar companies, policies and strategies that are built around “getting rich quick” can be risky. To fully understand the potential consequences of entering into an accelerator program or utilizing the accelerator model as a policy tool, there is an acute need for authoritative, robust data and metrics. Moreover, while any program can call itself an accelerator, it is important for entrepreneurs and policymakers to understand which programs actually accelerate startups and which ones do not. This distinction is especially important when comparing accelerators to incubator programs, which provide services similar to accelerators but are completely different entities.

Some current accelerator data sources lack reliability and authority.

Currently, there are multiple data sources for accelerator information but many are lacking in reliability and authority. Seed-DB is the best-known and most widely used database for information on accelerators. Seed-DB provides entrepreneurs, researchers, and accelerator programs with recent data on accelerator demographic and portfolio information. These data include information on the location of accelerators, their industries of focus, the number of companies in their portfolio, and the value of their portfolios.

Seed-DB relies on monthly data updates from Crunchbase, a database of startup investment information that relies on voluntary or “crowdsourced” input. While data sources are forthcoming about the biases and gaps implicit in this type of data collection, it does not nullify potential threats to validity. For example, Seed-DB is missing many key data fields for accelerators. As Figure 1 shows, almost 40 percent of the accelerator entries on Seed-DB lack information on either the number of startups in their portfolio or the financial value of their portfolios.

In some cases, as with the substantial number of startups missing exit funding data, data are missing because of the natural time lag when collecting information on startups since they are so young. However these missing data have an effect on the utility of the database and as a result bias many of the inferences that a researcher might be able to make. These biases heavily contribute to threats to validity in research using popular accelerator databases because they contain so many outliers. Given that the databases list vastly different programs side by side—accelerators that have relatively massive portfolios and are investing in companies that may have valuations in the millions of dollars alongside small accelerators with only a few small startups—many of the statistical measures generated from these databases can be expected to have a high variance. Similarly, depending on how these data are treated, estimates based on these data may be varying and unreliable. The uncertainty concerning accelerator programs and their effectiveness combined with this lack of data has generated a pressing need for robust and authoritative data and metrics on them.