The inconvenient truth of venture capital investment data

Mark Suster is asking “why the hell has seed financing declined so much in the past 3 years?” in a post that’s receiving much attention. Fred Wilson followed up with “The Seed Slump“.

That’s two very prominent VCs diving into a narrative that many others have taken as given as well; only few have challenged the premise. (Update: Mark Suster’s response to this post).

But here’s the thing: the seed investment data does not show what most people think.

Because of a reporting lag, the most recent seed investment data is systematically understated. The below chart demonstrates the real issue. Seed investment data keeps being restated upwards in new reports, as new information surfaces. For example, the number of seed rounds in 2015 was first reported as 3,990, but two years later, this number had reached 5,680. That’s an increase of 42%!

The point here is not just that seed investment data is incomplete. Rather, it’s that the last ~24 months of data will almost always show declines, which later largely disappear. Granted, the U.S. trend in recent years seems to go beyond that. So it could be that seed investing in the U.S. has in fact declined. But even so, it’s almost certainly not a slump.

Interpreting self-reported data

Venture capital investment data is all self-reported. That means there are few standards, little consistency, and no reporting deadlines. Fundraising announcements are essentially a marketing or signalling tool for the company, directed towards media, peers, customers, potential employees, and future investors. Seed investment data is especially hard to collect, as a result of these challenges. Often, seed rounds are only reported once the company completes certain milestones such as a Series A funding round, a key hire, product launch; whatever is needed to feel confident enough to go public.

Most of the U.S. commentators have been sourcing their data from Pitchbook/NVCA, which do an enviable job collecting this data directly from VCs and news. A lot of the self-reporting in the U.S. traditionally also happens via Crunchbase (and AngelList to some extent), which have been limiting API access towards third-party vendors. Moreover, there is also anecdotal evidence that U.S. founders have become less inclinded to self-report than before via these channels, for various reasons.

Establishing a more meaningful interpretation

At Dealroom, we face the exact same reporting lag issue with the data we collect ourselves. But instead of taking the data as reported at face value, we challenged ourselves to critique our own numbers and try to get closer to the truth. As a result, we were able to find some clues what the reality might look like, at least directionally.

In a report called The Journey to Series A in Europe which we published in October 2018 in partnership with Atomico and LocalGlobe, we have analysed European seed investment data in detail. The dataset is based on Dealroom’s own database but with two important modifications.

Report - The Journey to Series A in Europe (with Atomico and LocalGlobe)

First, we analysed the number of months between when seed rounds actually happened and when they were reported. Based on this estimated reporting lag we have estimated the “adjusted” number of seed rounds in the last 18 months.

Second, we applied a systematic relabelling of more than 22,000 European funding rounds based on round size and sequence. In summary:

  • Seed is the first round to be >=$1m
  • Series A is the first round to be >=$4m and <$15m, split between Old Series A (<=$7m) and New Series A (>$7m)
  • Any round before the Seed that is >=$0.25m is labelled as a Pre-Seed round
  • Any round between Seed and Series A is labelled as a Bridge round

Distinguishing pre-seed from seed is especially relevant in this context. The reporting lag is most severe in pre-seed: more than half of all pre-seed rounds are reported only much later (or never, we must unfortunately realise).

The below chart shows the end-result. Instead of a 35% decline from 730 to 460 rounds between 2016 and 2018, the adjusted data shows an increase of 23%.

This method isn’t perfect of course, but at least it’s less inaccurate.

What does this say about U.S. Seed investing?

The U.S. Seed slump is certainly an exaggeration. But the argument that the funnel from Seed to Series A has steepened in the U.S. probably remains valid. And the original post “Is VC still a thing?” is an absolute must read.

The Journey to Series A in Europe

Back in Europe, venture capital is in a very different state of development. Both traditional VC and Seed investing have been growing at a more similar pace.

The full report itself addresses a broader range of questions faced by European startup founders. For instance, only 19% of European startups reach Series A, within 3 years of doing their seed round. The report shows what founders can do to increase their chances, including which firms to raise from.

It would be interesting to see what Pitchbook/NVCA’s data would look like after applying a similar methodology. We would welcome the opportunity to share our methodology in more detail.

Report - The Journey to Series A in Europe (with Atomico and LocalGlobe)