Data Activation in Venture Capital | Dawn Capital

Charting the evolution of data engineering strategy in prominent VC firm

At the heart of the newsletter is a commitment to exploring the dynamic field of data activation,  i.e. how to capitalize the data stored in your data warehouse. My journey so far has predominantly revolved around the B2B SaaS world, where I immersed myself over the past couple years. However, today's edition marks a shift towards a realm I've always been passionate about: venture capital.

VC firms are not typically known for setting up extensive data centers or directly acquiring cutting-edge hardware to run their operations. Instead, they operate in a domain where nuanced, 'small data' can have a transformative impact. I am a firm believer in viewing VC through the lens of a product, with data playing a critical role in its foundational structure.

Over the past few years, I've delved into how VC firms harness data and the significant value they derive from it. Today's modern VC landscape features firms that are increasingly attuned to the data environment. While they may not always be tech-first organizations, some have distinguished themselves by developing sophisticated technology that enhances various operational aspects from startup sourcing to due diligence. 

In this series, we're excited to welcome Ties Boukema from Dawn Capital. Dawn manages five funds, along with three additional opportunity funds, to continue backing the portfolio winners. Dawn is not just participating in the VC industry; they are actively shaping it by embracing data-driven strategies. Ties brings a wealth of experience in data and consulting to the table. Having spent the last five years at Google, he was instrumental in overseeing all business intelligence, trading, and revenue reporting for the Enterprise Ads solutions in the EMEA region, covering multiple billions of dollars of revenue.

I joined VC rather than going to another tech company, because I felt there was a massive opportunity for Data & AI to improve dealflow and returns.
The closer you are to the investment side at a big VC the closer you are to where the magic happens. And so that has always been the intention.

I caught up with Ties for our blog interview. Let's dive in!

VC as a Product

Most of the attention for data & AI in VC goes to finding early stage deals. And as a pre-seed/seed focussed fund, if you are not leveraging data properly to scout & understand the landscape, you are trying to find needles in a haystack by hand (rather than using a metal detector).

For later stage funds, in private markets you will have to rely on your network & relationship much more. This doesn’t mean however that data & AI are not hugely valuable.

The most exciting - and overlooked - opportunity for me is around relationship intelligence for later stage funds. If before every VC-to-VC meeting, we have all relevant intel at our fingertips, we will get more out of those meetings. We also can become much smarter about who we are meeting with and when through tech. Finally, we can spot ways to access deals in a way you couldn’t otherwise. If we are getting in touch with a startup, knowing that the CTO there is a former CTO of a Dawn company is hugely powerful.

Relationship building

At Dawn, Ties’s team has been instrumental in leveraging the data function to transform how the firm builds and maintains relationships within the venture capital ecosystem. A casual interaction among VCs, such as coffee meetings, where significant opportunities for relationship building occur is often overlooked by operational software. Despite the frequency of these informal meetings, there's a lack of technological augmentation to make them more productive. VCs often find themselves in numerous meetings throughout the day, with limited time to prepare or research each interaction thoroughly. 

Rolodex (as the relationship intelligence tech stack is called internally) summarizes key intel just before a meeting, arming the VC with up-to-date insights on the other party’s. This transforms these brief interactions into opportunities for deeper engagement, demonstrating the fund's commitment and understanding of their counterparts' ventures.

Data Team's Role in VC

Every fund has its unique aspects, but there are common elements shared among VCs, especially in larger funds. Dawn, being a $2 billion fund, involves extensive reporting both internally and to LPs. This reporting is crucial for us to grasp how companies are performing and presents an opportunity for automation.

Secondly, a good data team provides a different perspective. One of the projects Ties did first when landing the job at Dawn was “Strike Zone”, where the team created benchmarks & analyses of how different profiles of deals return. For instance, what is your MOIC on an initial check vs all your follow-on investment? This is something that influences fund construction so you can be smarter about where you put money.

Metrics 

Coming from a B2B SaaS background, we often have different strokes for calculating crucial metrics like CAC and ARR. Like every non-GAAP, there's room for interpretation, but transparency and consistency are key. I was curious how Dawn reconciles these varied approaches to metric calculation. Daniil Chepenko

Rather than imposing a one-size-fits-all model, Dawn adapts its strategies to suit the unique needs of each company, especially considering their stage and business model. For instance, in their SaaS investments, a PLG company would have different key performance indicators compared to a sales-led company with a limited number of clients. The funding stage also matters a lot here. For example, the significance of Customer Acquisition Cost (CAC) can vary greatly between direct-run businesses and those operating under subscription-based software models with Average Contract Values (ACVs). Sometimes, CAC, is not even relevant or available, especially for early-stage companies, so the fund opts not to burden the founders with unnecessary data requests.  This approach not only makes sense from an investment standpoint but also aligns with the companies' internal reporting, as these metrics are already central to their operations. 

I’ve heard some horror stories from founders that get asked by VCs “what's the Gross Margin for the French/Spanish/Dutch client base?” when that represents <10% of the overall revenue. The admin is not worth the insight in those cases.

Ties also emphasized asking for the raw data rather than pre-calculated metrics where possible. allowing them to apply their own standardized calculations for consistency across their portfolio. This method eases the reporting process for startups while allowing Dawn Capital to maintain accuracy and comparability.

Team Dynamics and the first data hire in VC fund

Fundamentally the data function at Dawn exists to help find & close better deals. It is a revenue generating function. Some elements like helping to automate internal reporting saved so much time for the investment team that they are worth doing as a one-off, but all other work is evaluated by its contribution to dealflow.

I think your first data hire should be someone with a real data engineering background. Even working with Crunchbase & Affinity APIs is much more of a pain than expected, and if someone cannot do the work the vision is worthless.
At the same time, you need to be able to sell the vision to the partners in a way that excites them.If your work does not move the needle for the front office, you are always a back office function.

Experiments with LLMs (Large Language Models):

The precision and speed of LLMs are paramount, and can definitely streamline tedious tasks like preparing slide presentations. But even near-perfect data accuracy is insufficient for financial reporting – it’s not sufficient that 95% of the data is correct.

In Rolodex, Dawn leverages LLMs to synthesize insights and create outputs that are easy to interpret for our relationship intelligence data products. Again, given how core relationships are for how we run our fund, leveraging AI here has been the highest impact use case so far. Data Engineer, Gavin, has been instrumental in bringing this to life.

We are also experimenting actively with ways to use LLMs to do operational work faster and better. When you do 15 expert calls on a company, you can use LLMS to synthesize key findings and quotes across them, and even flag any discrepancies with previous expert calls in the same industry. This is not AI helping you work faster - it’s AI helping you work better.