10 Mistakes to Avoid on Your AI in VC Journey
Learnings From Digitizing a 27 Year Old Investment Firm
👋 Hi, I’m Andre and welcome to my weekly newsletter, Data-driven VC. Every Tuesday, I publish “Insights” to digest the most relevant startup research & reports, and every Thursday, I publish “Essays” that cover hands-on insights about data-driven innovation & AI in VC. Follow along to understand how startup investing becomes more data-driven, why it matters, and what it means for you.
Current subscribers: 20,660, +256 since last week
Brought to you by VESTBERRY - Portfolio Intelligence Platform for data-driven VCs
Your portfolio data is a powerful source of insights. However, accessing these insights is a challenging task. Putting together the right metrics and graphs can take hours. Could you have one dashboard that lets you quickly and easily analyze your data?
Watch this video and learn how to transform your portfolio data into visual insights with just one dashboard. Whether you need an ad hoc analysis or an LP asks you about the performance of their investment, this dashboard has you covered.
I just got back from the “GenAI Wednesday” event in Munich where I gave a talk about “Data-driven innovation & AI in VC”. One of the most frequent questions - not only at the event but also more generally - was where to start your data-driven journey.
As I’m writing this article here burning the midnight oil, I decided to dedicate today’s episode to some lessons learned and 10 mistakes to avoid on your AI in VC journey.
1. Join a community of like-minded peers. You’re not alone.
When I embarked on the journey of making venture capital more data-driven in 2017, I felt lonely. Everyone told me “it’s not possible” and “others have tried” but nobody - except my partners - encouraged me to push for change. The industry was just not ready.
Fortunately, I trusted my instincts, similar to a few other lonely soles I met during my PhD and first years as an investor. We began connecting, exchanging our struggles, brainstorming, and contemplating the future of venture capital through one-on-one discussions. Gradually, these conversations evolved into small WhatsApp and Slack groups, which have been expanding steadily over time.
With ChatGPT and mass awareness for AI adoption, things started to accelerate end of 2022. Suddenly everyone and his dog was looking to become more data-driven and leverage AI. I started receiving numerous questions from other VCs and got invited to many more “AI in VC”, “Data in VC”, “Datahunt”, “VC <> Data”, and similar groups - majority of which became ghost towns quickly as most people wanted to take but few wanted to give. Zooming out, these groups are just too fragmented and not actively managed.
Counting the members across all groups I’m in, I get to around 300. This is in contrast to 20.6k subscribers of Data-driven VC. What about the other 20.3k? Right, you feel left alone with your legacy tool stack and pressure to become more efficient. But you are not. You are part of the big majority that feels lost in the noise. I can see it in the tens and hundreds of DMs I receive every week.
Unfortunately, I cannot answer all messages but I decided to build something more scalable, something better than the fragmented ghost towns: A community platform for everyone seriously interested in data-driven innovation and AI in venture capital. For those who are willing to give and share, and not just take and hide. Those who want to become stronger together, while at the same time keeping your secret sauce and winning an edge.
This is what a community is all about and I will make sure you’ll find everything to run a modern VC firm in one place: Highly actionable content, playbooks and best practices, podcasts and video recordings, tech stacks of similar firms, tool reviews and discounts, database benchmarkings, and most importantly, moderated community channels.
Sign up below to join the waitlist for our Data-driven VC community. There is no reason to be alone.
2. Get top-down support from your GPs.
Once you surround yourself with like-minded peers, you will quickly understand where the industry is heading and what you can do to catch up. Thereafter, you need buy-in from your GPs as any investment in tools and engineering HR will cut from their profits.
So better have a convincing story as the lack of top-down buy-in makes it a no-starter. Make sure you’re all on the same page. How much are you willing to spend? What are the expectations?
3. Laser focus: Define what you actually want.
Many start buying datasets and comparing CRMs without actually knowing what they want. Don’t be one of them. Take a step back and define your firm’s strengths and weaknesses. What works well and what can be improved?
While for some it’s increased deal flow and coverage, it’s better prioritization and access for others. Define your goals and be critical about which initiatives contribute to your North Star. Don’t get lost in the noise and just try to copy what others have done. Even worse, don’t be reactive to the masses of tool providers reaching out. Be proactive, do your home work, know what you want.
4. Measure what matters to ensure you’re on track.
Once you’ve defined your goals, make sure to introduce metrics to track your progress. What is your Northstar Metric? For example, if you decide you want to increase deal flow coverage, your metric should be something like hit rate or miss rate, as described in “Measure What Matters to Improve Coverage and Performance”
Better prioritization <> funnel conversion rates. Better access <> outreach response rates. Better deal winning <> term sheet acceptance rate. Better portfolio value creation <> founder NPS. You get the point.
VC has long feedback cycles and we cannot afford to wait a decade before knowing whether our initiative is on track. Without the right operational metrics, you’ll be tapping into the dark and risk losing the trust of your internal sponsors, both from the investors/users and the GPs paying the party (see 2. above).
5. Your 1st Hire: A Product Lead who sufficiently understands both worlds - Investment and Tech.
Now that you know what is possible, have internal buy-in, budget, defined your goals, and know how to measure what matters, you can finally move towards implementation. If you’re reading this, the person in charge is most likely yourself. It was the same for me and between 2017 and 2020 I built the whole tech stack for myself. Developer and user all in one.
While this initially seemed like a bottleneck, it was the best that could happen to me. Firms that hired their first engineer too early missed one critical aspect: The person in charge needs to be someone who understands both worlds - investment and tech. If this is not you, then hire someone who fills this requirement or alternatively hire an engineer and have her shadow the investment team for several months.
It’s critical to understand the investment process end-to-end before getting into implementation. Only that way you’ll be able to take the next steps such as hiring further team members, translating between the investment and engineering team, defining the product roadmap, deciding on make vs buy, and more.
6. Beware of hiring deep experts too early; scrappiness and hustle > seniority and expertise
This is a controversial one and I’m sure not everyone will agree but from my experience of having hired numerous freelancers and internal engineering profiles over the years, I’m convinced that it’s more about intrinsics than experience.
Digitizing an investment firm is in the first place about understanding the existing processes, defining the target state, and translating this into technical requirements. As fancy as all the AI and data stuff might sound, however, it requires more of a scrappy 80/20 hustler mentality than 100% deep expertise.
Find a Swiss army knife. Someone who has a couple of years of experience developing full-stack data applications. From collection over ingestion to processing and displaying. You don’t need this ivory tower NLP PhD with 5 A+ papers. Rather the opposite, you need someone who wants to get her hands dirty.
7. Make vs buy: Avoid reinventing the wheel
As you know what you want, think twice before re-inventing the wheel. Is it available? How critical is this component for your core business? Nice to have or a long-term moat?
I’ve written about the “Rise of Investment Tech” before and different from when I started, you can find a broad range of off-the-shelf tools today. So if it’s available and not core to your business: buy it. If it’s available but core to your business: build it. And if it’s not available and .. well, you build it ;)
Today, access to public data at scale might still be a moat, but tomorrow it will be a commodity. The combination of public and internal, sensitive private data, however, is what will create a long-term competitive advantage. Moreover, the signals that you generate from processing will be crucial.
8. Push vs pull: Getting the right user culture
There are three buckets of blockers: 1) lack of budget and top-down buy-in, 2) technical feasibility, which contains finding, selecting, and attracting the right engineering talent, being smart about make vs buy, and implementing, and lastly 3) the wrong culture and lack of bottom-up buy-in.
While 1) and 2) seem more critical than 3), I can tell that most firms have failed due to 3). Changing the culture in VC, a rusty old-school cottage industry that has seen close to zero change since its inception in the 1950s, is hard. Very hard.
It’s easier if you launch a firm with data & AI in its core from day 1, like Signalfire or Moonfire, but unfortunately, the majority of firms out there already exist without this DNA. So how do you transform the culture of your firm to convert your biggest critics, i.e. the Analysts and Associates who already hate to enter the notes into your CRM system, into raving fans of your data platform?
Well, unfortunately, I don’t have the silver bullet but I can assure you, you better find out early enough. On that note, I also heard from several other firms that the data leads launched their platforms too early (in the 0 to 1 manner) and their investment team members started to focus on the shortcomings instead of the wins, quickly losing trust and choking off the whole project.
In most cases, you’ll only have one shot. There is no way to replace your user group if the first attempt fails. Different to the startup playbook, it’s wise to take a bit longer before launching premature products to not lose trust and get it right with the first attempt.
9. One step at a time: Do less, but do it right
It’s easy to get overwhelmed by the data & AI noise. Even more important to stay focused, not lose sight, and take one step at a time. Know what you want, focus, iterate fast, and finish. Don’t start too many things at the same time, but limit yourself and prioritize accordingly. It’s better to get a few things really right versus many things half-baken.
10. Create redundancy: No single point of failure
As your initiatives come to fruition, zoom out from time to time and look at the big picture. Are you still on track? What does your North Star metric tell? If all lights on green, adoption of your platform is growing, more deals are coming in, coverage is increasing, access is getting better, efficiency is up, and the engine overall is just running well, zoom out and look for the points of failure.
If done right, this initiative will become core for your business sooner than you expect. Even more important to stay paranoid and identify points of failure early on. Is your product lead still happy? Your engineer got another offer? Database hacked and no copy? The list goes on, so be aware as the importance increases, you should create redundancy. Across the team as much as the stack itself.
Stay driven,
Andre
Thank you for reading this episode. If you enjoyed it, leave a like or comment, and share it with your friends. If you don’t like it, you can always update your preferences to receive just the regular Thursday “Essays”, just the Tuesday “Insights”, or both. Subscribe below and follow me on LinkedIn or Twitter to never miss data-driven VC updates again.
What do you think about my weekly Newsletter? Love it | It's great | Good | Okay-ish | Stop it
If you have any suggestions, want me to feature an article, research, your tech stack or list a job, hit me up! I would love to include it in my next edition😎
How would you do this differently if you were building it from First Fund onward?