What would happen if one would use the Twitter data of almost 600 Dutch startups to map their interaction into one large network? Look through the eyes of Dutch startups to the local ecosystem, and see which parties are important to them.
- 474.613 tweets and counting
- 598 Dutch startups
- 119.000 mentioned parties (see if you are included)
- Stabilized network: over last 5 years
- Graphs represent top mentioned users (see if you are included)
- 13 interviews (why you ask?)
General overview of top 365 mentioned users:
How to read such a network?
- Every node [dot] represents a Twitter user account
- If one user mentions another, this creates an edge [link] between the two nodes
- To more mentions, the stronger the traction between the nodes
- The node size represents the amount of times the user is mentioned
- The colours express statistical communities
What can be done with such a network analysis? This research zooms in to particular participants to see where they are positioned, with additional interviews to see how they tap into the entire network. Social entrepreneurs with growth / scaling potential are chosen because of their impact-first mentality over a sole revenue-based focus. Included in this profile:
Where are they positioned in the network above?
Another way to visual how these ‘social startup’ tie into the ecosystem is by colouring their connections black:
What can be said about the position and integration of the nine social startups?
- Social entrepreneurs are not one cluster, but spread over the network
- They have a strong reach, tapping into many communities
- The main theme around they are organized is the share economy / sustainable entrepreneur
- Connecting actors or bridge builders are important for their organisation and integration
Together with these social startup, another 5 organisations have been interviewed for their role seems important to integrated the share economy / sustainable entrepreneur cluster, being: Pakhuis de Zwijger, Sprout, Share NL, Social Enterprise NL and Impact Hub Amsterdam.
A cluster-based analysis
Last but not least, a general cluster-based analysis allows to find the various clusters distinguished in the Dutch Startup Ecosystem. Simply looking into the actors involved in such clusters (together with interview data – see below), indicates some themes around which the clusters organize themselves. A zoomable map of this analysis can be downloaded here.
But why these interviews? Talking to people on ground level is crucial to understand the mechanisms at work under the surface of these graphs. Interviews helped to explain, nuance and complicate the findings in the network while creating sensitivity to the limitations of Twitter-driven analysis too. See what implications these interviews lead to in my recent blogpost ‘the politics of network graphs‘. Some of the limitaitons can be found in the presentation I did at the Entrepreneurship Future conference.
- The critique that many of the expert-driven ecosystem analyses produce laundry lists is actually expressed by Stam (2014)*
- Throughout the use of earlier notions of an entrepreneurial ecosystem which derived from economic cluster analyses and system thinking, network theory has always been one of the explananda to understand an ecosystem. Drawing on the widely used biological metaphor, I operationalized an ecosystem as a heterogeneous set of interdependent actors.
- In my research I try to step away from the expert as the starting point, instead, I re-appropriate (non-evoked) Twitter data produced by Dutch entrepreneurs. In other words: I used the online Twitter activity of 598 Dutch startup entrepreneurs to map their affiliation network based on who they interact with (@mention).
- The sample of entrepreneurs was informed by the entrepreneurs database Dealroom, a combination of user-generated and aggregated information on startup enterprises. I bypassed the discussion on what would qualify a startup, by using their database only as an entry point into the field.
- In order to get to an interpretation of what the network represents, I ask various entrepreneurs that were included in the graph to interpret the graph, there position, and their surroundings. This is the step from an exploratory descriptive network to a more explanatory interpretation which I would argue is what makes it a valid entrepreneurial ecosystem approach. These collaborative readings allowed me to know the limitations of a twitter-driven network, and the implications of representing an ecosystem as a network (rather than a diagram).
* Stam, E. (2014). The Dutch entrepreneurial ecosystem. Available at SSRN 2473475.
Or drop a line at:
mail [@] jeroendevos [.] nl or +316 24 542 807