I come to accelerators, incubators, and innovation programming from two angles. Having been involved in the startup sector for years, I was interested in the development of accelerators, so back in 2014 I set up my own one. It was a specialist healthcare program focussing just on eye-care, called EyeFocus. It mixed elements of traditional commercial accelerators with corporate accelerators and social impact programs.
EyeFocus was backed by Bayer Healthcare and Zeiss, made investments, and ran a bootcamp program. It also supported some social enterprises and very early stage teams, and it focussed more than usual on building what we called ‘the eye-care innovation ecosystem.’ This meant that it not only focussed on making money, but also on creating value for the wider ecosystem,* working with non-profits, universities, and hospitals, and aiming to leave that ecosystem better connected as a result.
It was not directly a corporate venture program because it was owned and run by me, rather than being run by me as an employee of a corporate owner. However, it was focussed on connecting a leading pharma company and hardware company into the startup space to promote their engagement with innovators. In that sense, the aims were closely tied to corporate venture; searching for potential investments, and creating a two-way flow of information and contacts between the corporates and the wider innovation ecosystem around them.
In parallel with the business, EyeFocus was the subject of a PhD I wrote on what I call Applied Social Network Theory; a mixture of sociology and business. In it I analysed the underlying social network of a accelerators. The research examined things that myself and other accelerator founders know and do instinctively, explaining them by applying the language and concepts of Social Network Theory.
In particular I wanted to understand things like why mentors mentor, and why accelerators enjoy very high levels of trust. It seems just obvious, or the ‘way we do things,’ but it’s actually very counter-intuitive from a more corporate perspective that experts would give their time for free, and startups would share their precious new ideas so openly with strangers.
In the PhD I think I answered these two questions, finding that the trust and willingness to mentor stems from accelerators enjoying very strong social capital in their wider ecosystems. This enables them to use social validation to reward people who cooperate, and the threat of reputational damage to sanction those who breach their norms.
The PhD also looked at the sort of network structures that best lead to innovation, and how social capital is a currency for accelerators. The result is a number of findings about how better to design innovation programs. This applies to accelerators, but also to corporate venture programs, incubators, and other innovation programming.
Who this is for
The findings are relevant to people thinking about running accelerators and to those working on corporate venture. The social network aspect of designing innovation programs is challenging for large corporations or government agencies because their networks are most different to those that I find support incubation and innovation.
Large corporate networks are typically quite closed, with a lot of internal duplication of contacts, but fewer weak ties out into other networks. Corporates and government are also more risk averse, so are less likely to engage with people informally, without NDAs, and to operate the sort of open cultures you find in accelerators.
Some of the pointers below propose underlying social network structures and dynamics not typical for large corporates or government agencies. However, where the aim is to find new ideas, adopt new innovation practices and culture, and to promote out into new networks then weaker and more open social networks will support this, and closed ones will inhibit it.
Some ideas on innovation programming
Here is an overview of some of the insights and conclusions from the research, combined with my own experience running EyeFocus, and working with other accelerators. I will be writing up and posting more over the coming months.
1. Incubators and accelerators are different things!
People constantly mix up accelerators and incubators. They are different. Put simply, an accelerator is a program, and an incubator is a building. Accelerators usually have a cohort of people who join for a fixed-term, and provide value primarily through mentoring. They mainly do not charge the startups to be involved and make money through equity investing or sponsorship. Incubators are generally a building or space in which companies are offered desk space and business support in return for rent.
Both are forms of incubation, which is mainly support for early-stage businesses, but can extend to support for individuals, teams, and other entities generally engaged in doing something new. Accelerators can therefore run within incubators, but not the other way around. One is a program, the other is a building.
2. Accelerators are social networks
Accelerators, incubators, and other innovation programs are seen primarily as business support programs which happen to create social networks. I suggest that actually they are a specific type of social network, which result in business support as an outcome of their social network structure.
To understand this I broke down that network into its component parts to try to see what it looks like, what is unique about it, and why it supports innovation.
Accelerator social networks have some unique features in common. They consist of a tight core of people, mainly the accelerator team and the startups. These form a strong bond, and develop a culture. In Social Network Theory terms I describe this part of the network as featuring strong ties with high network closure, which has a high level of social capital. This part of the network is interconnected, with many overlapping ties (strong ties with high network closure). They spend a lot of time together, working and socialising, so they form close ties, high levels of trust, and benefit mutually when they succeed. The shared culture they develop forms the norms of the network — the rules, the culture, and the reward systems.
Around this core, when done properly, the accelerator builds a diverse weak tie network of mentors and other partners. Weak ties are people not well known to each other, who do not share many common contacts. The accelerator becomes embedded within this weak network architecture, where there are not a lot of overlapping connections, and which has many bridging ties.
Bridging ties are people who can bridge between different interconnected networks that are not already connected to each other, creating value by introducing a lot of ideas and people to each other. Not all weak ties are bridging ties, but most bridging ties are weak ties, so weak ties are important to access new information and contacts.
This network is a weak tie network because these people are (or should be) from a very diverse set of backgrounds and other networks. They create bridges from the dense core out to their own networks, and beyond. This enables the accelerator to support its startups by embedding them in a very big and diverse social network through which to find knowledge, expertise, people, investment, and routes to market.
Creating a diverse, weak tie network leads to an exponential effect in which the startups meet some tens of mentors, but are consequently connected by them directly into a wider network of thousands of people.
This is why good accelerators are open, always pulling in new people, and have large international networks. This is also why people who set up accelerators tend to be gregarious, well connected, and highly social. In effect, accelerators are highly complex social networks, rich in weak ties and social capital.
3. Social Capital as the currency of accelerators
There is an interesting exchange of value going on in this network structure, which is why (real) accelerators rarely pay mentors, for example. The value exchange sees the mentors bring value to the startups, on behalf of the accelerator, because the accelerator rewards the mentors with a range of value that is embedded within its network, namely:
- Social validation (being promoted as a mentor)
- Social ties (introductions to the other mentors, investors, experts)
- Access to novel information (startups with new technology or ideas)
I call these forms of value social capital* because they are distinct from economic capital (money), and because this value is embedded within a social network. If the accelerator does not consciously build the right social network structure then it will not have the social capital with which to reward its mentors.
When done well, an innovation program like an accelerator should be able to offer a lot of value to people in return for mentoring, or otherwise supporting it, and should not need to pay many people. Payment changes the dynamic and incentive, and turns people into consultants. It is not always bad, but should be reserved for programs that really cannot offer enough social capital to encourage people to mentor their startups.
When designing programs, it would be worthwhile to map where the social capital will be created, and how it will be deployed, in the same way as the economic capital is mapped in a P&L.
This explains the role accelerators play as well, suggesting that they accumulate value in the form of Social Capital as a consequence of being connectors within social networks, and then confer their Social Capital on the startups in their cohorts to support and protect them. These startups might otherwise have very little Social Capital of their own because they are invariably new and young. Without something to give back, they are in a weak position to ask for favours from those around them. By conferring its accumulated Social Capital on its cohort, the accelerator gives the startups a higher status within the network, and something with which to return favours. They are then equipped to ask for support from mentors and the other stakeholders in an accelerator.
4. Non-redundant novel information
A part of Social Network Theory that I found explained a lot of about accelerators is the concept of non-redundant information. It also explains in more detail the role of social capital. In a social network, information that is not easily sourced, and is new, has value attached to it. The information is non-redundant because it is new to those who receive it. For example, if I tell you it is raining and you should take your umbrella, that is useful information you value. If someone else then tells you it’s raining, that information is now redundant, and not valued by you.
This non-redundant information becomes a form of social currency, or social capital. In a network, if you have a lot of non-redundant information you are valuable to other people. However, once that information has been shared with a few people, who share it with a few more people, over time that non-redundancy decays. More and more people have already received the information, so it becomes redundant if they receive it a second or third time. Eventually it loses its value as a form of social capital.
Accelerators create a resource rich in non-redundant information. The cohort of startups, working on new and novel ideas, are novel, non-redundant information for investors and mentors. This incentivises them to engage with and support the accelerator. For the startups the mentors and investors are a large pool of information and contacts that is to them non-redundant. Hence, both groups are attracted to the accelerator, which for each is rich in non-redundant information. This explains why startups join accelerators, why investors back them, and why mentors support them.
Hence my argument that social capital is the currency of innovation programs like accelerators. They start out rich in non-redundant information that is novel to those involved, and promises people early and privileged access to this information in return for their cooperation. People cluster around the program and contribute towards it because they all benefit from early access to that non-redundant information.
5. Decay of novel information
At the beginning of a program, an accelerator is therefore rich in social capital because it is rich in non-redundant information. It is in a strong position to ask favours of the mentors (to mentor the startups), and the startups are also willing to part with time, equity, and other forms of value in return for the non-redundant information residing in the mentors and the wider network.
However, over time the mentors and investors will have met all the startups, and the startups will have accessed the non-redundant information residing in the mentors and wider network of the accelerator. Therefore, the social capital within the accelerator diminishes as the non-redundancy of the information decays.
In my research I found that this may partially explain why accelerators have fixed term programs, and why they tend to run for 3–6 months. There comes a point in time when the program has expended all its social capital and becomes progressively less useful to everyone. In effect, with a roughly fixed number of startups and mentors, there is also a fixed number of novel interactions that are possible before everyone has met each other.
At that point, the program tends to stop. The next program restarts the cycle, with a new set of startups, who meet the mentors for the first time, which restores the high degree of non-redundancy, and begins the next cycle of its gradual decay. Understanding this decay of non-redundancy should allow programs to understand how their nature will change over time, and factor that into their design and agenda, adapting how value is created as the program progresses, and deciding when the program should stop.
This decay of value also ties into how accelerators police their norms and protect their startups, using social capital.
6. Trust and ‘link reciprocity’
It is interesting and unusual in the wider business world that (real) accelerators rarely use NDAs. Mentors are rarely contracted or asked to sign NDAs. There are various reasons for this, the primary one being that it is expensive, fairly pointless, and an obstacle to the free exchange of information between all the people around the accelerator program.
I looked at how the underlying social network supported this. I found that the social network dynamic that describes this is called link reciprocity. Very simply, a dynamic network that is constantly refreshing its members, (as opposed to the network being fixed and static), is in a position quickly to pull people in or cut them off. When people are trustworthy and supportive, they are rewarded by being pulled into the network. When they breach the norms of that network they are sanctioned by being cut out of it.
If the network is a repository of novel, non-redundant information and social ties, two key forms of value in a network, there is a cost for people whose ties are cut out of it, and a benefit for people who are welcomed into it.
Accelerators reward cooperators with early access to non-redundant information, which is perceived of as valuable. And they also reward them with social validation, because they can spread reputation widely through their weak tie networks.
This enables the accelerator not only to reward co-operators, but also to police its norms by threatening to cut off the flow to that non-redundant information and to affect reputation negatively through its large social network as a form of sanction towards bad actors.
For link reciprocity to work, there needs to be value embedded within the network, and it needs to be dynamic so people can be quickly pulled in and pushed out as part of the daily fluctuation in the network structure.
Consequently, it is possible to rely on a high degree of trust within a (real) accelerator network because it is rich in the forms of social capital represented by its non-redundant information (startups) and weak ties (mentors). It is also positioned to validate or damage reputation in a way that a single startup could not do alone.
If this is not understood, and mentors are asked to sign NDAs, firstly the people who are the most naturally valuable mentors are put off joining, and secondly the sanction used by the accelerator changes from social sanction to legal sanction. The personal cost of being cut out of the network is far higher than the risk of successfully being sued for breaching an NDA, so NDAs actually make the startups and program more vulnerable, not less.
The decay of non-redundancy in a program is also intrinsically tied to link reciprocity as it progressively reduces the program’s social capital, meaning it becomes progressively harder for the accelerator to enforce its social norms because the threat of sanction diminishes with the decay of the non-redundant information that is a main source of its social capital.
Understanding link reciprocity, and consciously structuring a dynamic network with a weak network architecture that is rich in non-redundancy should enable accelerators to better protect their startups, and incentivise mentors more effectively.
7. Cohorts, fixed terms, efficiency
Traditional accelerators have a number of features in common. These include an application process that selects a cohort of startups, and a fixed term program. In my research I found there are a number of reasons why these things matter and have evolved as they have.
The selection process means that the startups on a program have been validated, and represent the best in class. This forms the novel, non-redundant information that is used to reward mentors, and support link reciprocity.
Secondly, the cohort allows the accelerator to deliver great efficiency to all the various participants. For the startups, the cohort is like a class in a university, allowing value, such as training, to be delivered once to a class, rather than repeatedly to each individual. But the cohort also offers efficiency to mentors and investors, who can come and meet a validated group of startups in one go, rather than having to organise multiple meetings with each startup separately. By curating both the cohort and the mentor network, the accelerator performs a function for the ecosystem, expending effort on their behalf so they can connect efficiently with minimal coordination effort.
The fixed-term of a program is also tied to the cohort. If you have a cohort then they need to start at the same time, and follow a common program agenda.
Consequently, an accelerator, or similar innovation program is really about doing the work for other people so it can deliver value in a highly efficient manner.
8. Tacit and acquired knowledge
A large part of what accelerators do is knowledge transfer. Mentors support accelerators by sharing their knowledge and networks with the startups. Accelerators also offer training and workshops as part of a curriculum. In doing so, accelerators and innovation programs transfer two different types of knowledge, tacit and acquired knowledge. Tacit knowledge is built up over time, through experience. Tacit knowledge is more akin to wisdom. It cannot be taught, but can be transferred. Acquired knowledge can be learned, and therefore taught. This can be skills, such as accounting, or techniques like Lean Startup.
Accelerators transfer these two forms of knowledge in two distinct ways. Acquired knowledge is taught as part of a curriculum, by trainers and experts, or by mentors in workshops. Tacit knowledge is transferred through mentoring, where mentors discuss topics and ideas, share the lessons they learned over time and through experience.
It is important to recognise that innovation programming performs these two functions, and that training and mentoring achieve different outcomes.
9. Measuring success
Traditionally, the success of innovation programs like accelerators are evaluated in a relatively narrow way. The public sector looks for further investment raised and jobs created, and the private sector looks at investment outcomes. These are fine, but miss a large part of the bigger picture.
Firstly, I found with EyeFocus, and in my research, that a good accelerator that understands its role as a social network, in particular as an ecosystem builder, should be creating outcomes that are hard to measure, and don’t directly benefit it straight away. Creating a large, well networked, dynamic ecosystem in which the accelerator sits in a prominent central position makes it more likely that its startups will succeed because they are immersed in a healthier business environment.
Research has shown that when accelerators come to a city, not only does investment in the accelerated startups increase, but investment in startups across that city increases. This is because the accelerator attracts investors, and promotes investment, and when they visit they meet other startups, not just those in the program. Not all of those outcomes directly benefit the accelerator, but they create a better environment for all startups in the ecosystem.
Secondly, when viewed as social networks rather than business support programs one starts to think about the people on a program, rather than the startups. Most startups on accelerator programs consist of 1–3 people, so a program with 10 companies could have 20–30 people pass through its program. That is 20–30 people who leave better educated, with stronger social networks, and more complex social skills.
Whilst only a fraction of the startups might succeed into the future, those people will all go onto other things. These are rarely measured, but should be seen as equally valid outcomes. Measuring these outcomes is as important as counting how many companies raised more money and ‘succeeded.’
10. Some quick conclusions
When designing an innovation program it is important to understand the social network that is also being structured, and how it will affect the dynamics and outcomes of that program.
That network needs to have a core of startups and the accelerator team who share values, trust each other, spend social time together, and share their networks. This forms the core that produces the norms, or culture of the program. Around this there needs to be a large, diverse weak tie network of people not all connected with each other or with the startups and programs. They bring tacit knowledge and bridging ties out to other networks, and form a core of the value an accelerator or innovation program offers to the startups or participants.
If a corporation or government agency establishes an accelerator or innovation program, it is therefore important they focus on providing mentors who are not all from within their organisation, as that is not a weak tie network and will be of limited value.
> Rewarding Mentors
Mentors are rewarded with social capital, which consists of both the ability of the accelerator to provide early access to non-redundant information, but also social validation of the mentor by association with the program. It is therefore important that innovation programs make sure they focus on providing that value to mentors. If they choose to pay mentors they change the nature of the relationship and weaken their ability to enforce norms, protect their startups, and get the most value from those mentors. A certain type of person, who gets this, will mentor as a volunteer. Someone who wants to be paid to mentor has different motivations and is a business advisor or consultant. Both are valid, but they are different.
Accelerators therefore need to concentrate on how to make sure their mentors and other supporters receive sufficient social validation, social capital, and access to non-redundant information. This includes profiling them on websites and in newsletters, inviting them to events, and ensuring the wider value that the program enjoys is shared with them. Mentors should have plenty of opportunities to network with each other, and with other high-value contacts of the accelerator, not just with the startups.
> Open spaces
When choosing or designing a building to house an incubation program, the desired social network should be factored into the physical layout and design. A building with tight security, locked doors, and closed offices will restrict the social outcomes, whilst being as open as possible, with ample communal spaces will support network interactions. Also, allowing for social engagements such as drinks and meals supports the development of social capital outcomes.
> Location is important
Research shows that people are willing to travel far to attend accelerators but do not go far for an incubator. Incubators therefore tend to attract local businesses, or businesses that will travel to be near a specific source of expertise or market activity. Accelerators, being shorter fixed term programs, are easier for people to travel to, and then leave again. Accelerators need to be in ecosystems as they require a very large network of people around them, but incubators are more focussed on their internal network, and surrounding specific resources, so can be further from large and active ecosystems.
An accelerator running anywhere other than a major city or tech ecosystem may struggle to find a diverse group of mentors, and if there is a focus is very likely to struggle to find investors. That is not to say that such accelerators will fail or should not be embarked upon, but that these potential problems should be planned for from the outset. Solutions can involve using online mentoring more, and planning group visits to the nearest major hub.
> Fixed-term Programs
Accelerators need to be fixed-term with cohorts to create the efficiencies that make them valuable to everyone involved. Without cohorts and fixed terms they are effectively incubators. This creates different value: you could argue that incubators create a safe environment to nurture startups, accelerators give them rocket fuel so they either soar or crash. Both have a role in the wider process of incubation but they are different, and should be applied in different situations.
> Protecting Startups
Startups are typically quite vulnerable. They lack both the economic and social resources to protect themselves. Accelerators have a role in protecting startups, not just accelerating them. They do this by accumulating social capital, and being able to threaten sanction more powerfully than a startup. I’ve explained how social capital, weak network architecture, and link reciprocity underpin this function. It is important that accelertor-like programs understand that this is also their role, and that the network they build will affect how well they can do this.
> Investing in startups
Accelerators were initially a way to invest in a large number of early stage companies efficiently. They have evolved well beyond just that. It may be appropriate to follow that early model, but it should not be a given. Some accelerators will achieve more if they do not blanket invest in the cohort, especially when the aim is to build an ecosystem or scout a sector. Investing requires rejecting most applicants, and selecting a cohort primarily around whether they are good investments. This is a very narrow filter, and may work against a program’s wider aims. Other approaches can include investing half way through, or towards the end of the program, only in a few of the cohort who meet that specific set of requirements.
> Measuring outcomes
Investing can also mess up the evaluation of outcomes. Unless the primary aim of a program is investment returns, these should not be the main measure of success. Outcomes can include the personal development of people on the program, scouting a sector or technology, championing entrepreneurship, developing an ecosystem, or just finding and filtering a lot of technologies for a bigger project. The evaluation of programs should closely fit the aims for which it was set up. Startups on a program that ‘fail’ may also be successful outcomes, as those people may have joined other successful startups on the program, or may just leave better educated and networked and go on to other successful outcomes.
To measure the success of a program, look at the bigger picture. Look at all the people who were involved, at the ecosystem around the program, and the data, information, knowledge it generated. Share outcomes widely across the ecosystem for more successes.
You can read a review of literature that explains Social Network Theory here.
* I like the word ecosystem. Some people don’t. I feel it reflects in the business environment the parallel it draws with nature. An ecosystem is different to a network. Networks consist of people (nodes). Ecosystems consist of networks, but also other things like infrastructure, policy, and environment.
- * Social Capital is notoriously hard to define, which is why I’ve said this is ‘my’ definiton in this context.