Venture capital is often called an apprenticeship business in large part because experience matters and takes time to accumulate. But successful firms are able to translate and transfer experiential wisdom through institutional memory, which involves codifying what works and what doesn’t. Back in 2007, we did this with our life science team by pulling together a detailed list of “Lessons Learned” from our existing biotech portfolio. We recently went back and revisited that list of reflections. Since 2000, we’ve invested in nearly 60 biotech companies, so we’ve had a reasonable ‘n’ to think about and inform our collective observations.
It’s hard to judge how much we really learned from the 2007 process and applied to our deals since then, but it’s fair to say we’ve evolved our strategy around a number of these observations. Like all learning loops, it’s been a process of building the institutional memory around what’s generating returns, and appreciating the mistakes of the past. As my friend and coinvestor Kyle Lefkoff of Boulder Ventures often says: “let us make some original mistakes” – trying not to fall into the same troubles that hurt returns in the past.
Below is a very distilled and rather sanitized summary of our “Lessons Learned”; at the macro level, many of these are rather prosaic and not very insightful, but practical reality of each deal is where insight becomes actionable (and inaction causes issues). To spare the innocent, I’ve dropped most of the company names, but all these points were largely informed by experience in the trenches not abstract thinking.
Here are five buckets of reflections:
1. Management, management, management. Its not news to anyone that the success of a startup biotech depends in large part on the management team. This is an axiom in venture: getting the right group of early entrepreneurs and executives around the table is critical. But this is often not easy in early stage biotech companies.
- Different management teams are often required at different stages of a biotech, and the reality is that many seed- and early-stage deals don’t need a CEO. They are science-driven companies that need great Chief Scientific Officers to build the fundamentals of the story, and a BD executive to help build the broader vision. It’s upon that progress with which a company can recruit a great CEO. Putting a founding CSO in as the CEO early on often creates unnecessary conflict in the future: having the conversation about a perceived “demotion” to CSO when hiring the future CEO is uncomfortable and avoidable. Same goes for putting the lead BD entrepreneur in as CEO early on to “fill the role”. Keeping the role vacant in the beginning prevents future discomfort, or at least minimizes it
- With weak management, Boards often begin to run companies. And then it’s a vicious cycle: At the 1Q Board meeting, the management thinks they are responding to the Board so they chase after XYZ; then at the 2Q meeting, the Board says they think chasing ABC is a better idea, which the management does thinking its being responsive; and then at the 3Q meeting, the Board wonders why the company has no direction and chaos ensues. Never a good cycle, but of course its not as simple as this. A good Board is able to provide direction, governance, and input, and a good management is able to distill that feedback and integrate it into the strategic direction of the company. It’s a healthy balance and tension. But keeping the Board away from whiplashing the “day to day” program choices of what to “chase” is key. An important nuance is worth mentioning here though: an active Board Chairman or single Lead investor playing the role of an Acting CEO is very typical in an early stage startup, and is a good thing (especially given the point above about not having a formal CEO in a science-led startup at the beginning).
- Making management changes quickly is almost always the right answer. We have historically not moved fast enough to make senior management changes even when we knew it wasn’t working. Trusting one’s instincts is important: if it feels like its not working, it probably isn’t. And the team working in the company probably sees the same thing from their view of the executives. Further, if real management questions are present at the closing of a new investment its unlikely to improve: its often worth being explicit about this with the existing team before the closing to lay out expectations and possible action plans. Closing the deal and then firing the CEO immediately after doesn’t feel like the high road.
- Sometimes we’ve delayed for apparently good reasons (e.g., “we were close to a BD deal and didn’t want to rock the boat during the process”), which more often than not turns out to be poor reasons (e.g., “the deal never got done”). Ripping the bandaid off quickly is almost always the right answer.
- Don’t hire for the resume, hire for real talent. Lots of folks in biotech have good resumes, were part of stories with great drugs or great exits but didn’t actually shape them (or at least no where near as much as they think they shaped them). We’ve certainly hired our fair share of great paper-resume CEOs that didn’t translate into excellent leaders and operators in our startups. Diligencing the specifics of their actual contributions in the past is an important part of reference checking and recruiting. Some of the disconnect between paper and practice is that the transition from large to small companies is hard despite great past roles; others is that their past success was more luck than skill and repeating luck is a challenge.
2. You can’t pick your family, but you can pick your co-investors. The vast majority of biotech deals require >$15M in venture capital to get to an exit, and the average is close to $60M. This means that most deals require syndicates of at least three VCs and often more. This creates a lot of potential for entropy. Our observations:
- Big syndicates are often dysfunctional, and we try to avoid them. The bigger the syndicate, the more cooks in the kitchen, the more distracting the differing priorities become: different exit preferences (“go long” vs “sell early”), different capital appetites (capital-sparing vs pushing money into a deal), different views of a management team (change vs no change), different disease area interests (one made money on eye-deals, another lost money on eye-deals). Unless there’s a real lead investor and well-vetted group of like-minded partners, the big venture syndicates are frequently a struggle to manage. This is not always the case (e.g., Avila had five VCs and was a well-behaved Board), but it’s frequently an issue and becomes increasingly problematic as a company raises larger amounts of equity capital from broader sets of investors. One of the reasons we get involved at the seed-stage is so that we can shape the co-investors we work with.
- Misaligned investors can cause as much harm as weak management. This is undoubtedly true, and, for the reasons stated above, big syndicates increase the odds of misalignment. The best management teams work hard to keep their investor syndicates aligned and maintain a continual open Board dialogue about that alignment on strategy, vision, financing, exit, etc…
- Experience beyond the venture skillset can add real value on early stage boards. We are big believers that bringing great independent directors into deals early can be a huge help for startups. For example, getting former Vertex President and head of R&D Vicki Sato involved at Nimbus, even when the board was larger than the company, was the right thing to do and a great addition.
3. Diligence would benefit from more realistic crystal balls. The best way to not lose money in biotech is to not invest in biotech. It’s not a simple business, and deep due diligence around the science, programs, team, plan, patents, etc is critically important.
- Most drug discovery companies fail to deliver on their overly-optimistic initial Series A plan, so factor that into the financing plan. Almost every new discovery-stage startup comes in with a pitch that says “in 30 months we’ll have an IND”. Unless you have your lead scaffold in hand now, you’re not likely to get there in 30 months: novel modalities or new scaffold identification and optimization just take time. In our experience, it takes between 36-48 months to get to an IND around novel chemistry when the plan hits reality. Even a superbly executed plan can fall short: Avila’s BTK program filed its IND 42 months from the start of the company, 12 months later than forecast in the Series A plan. Factoring in the almost certain delay and slippage is important when thinking through capital and time requirements. The Catch-22 here is that if an entrepreneur pitches a plan for 48 months to an IND, even to an early stage investor like Atlas, its likely to receive a lukewarm reception; this leads to overly optimistic plans from everyone that 30 months can do it. It’s very healthy to have a frank conversation about this paradox and the issues around it early in a diligence process.
- Do your own primary diligence. Often when syndicates form around a new deal, the proverbial summer pool effect happens: since every parent thinks other parents are diligently watching the kids, no one does and bad things happen. Just because great VC firm XYZ is committing to do the deal doesn’t mean they did great diligence. We’ve learned the hard way that different firms, and different partners in the same firm, often do varying degrees of diligence. Sharing consultants and experts is fine, but having a direct dialogue with them is important: taking a fellow VC’s word for it that “former head of R&D John Doe thinks this is the hottest program ever” is simply not sufficient.
4. Super models and investment models: beauty is in the eye of the beholder. Lots of different investment models work in biotech, and most firms have their favorite. Some firms have done well with PIPES and later stage assets, like Longitude and Sofinnova, but it’s just not a space we’ve engaged in given our early stage innovation bias and the desire to keep doing what works for generating returns for us. Our experience with different models over the past decade has led to a few lessons for us:
- Going in early allows us to shape the DNA of the company. Whether it turns out to be the bleeding edge or leading edge, some of the best deals in our portfolio over time have come from the roll-up-your-sleeves model of venture creation. This is a recurrent theme on this blog, so I’ll leave it at that.
- If you can tranche the capital into a deal around important milestones, you should. By metering in the money, an investor can monitor not only the derisking of the science, but equally important we can observe how the team executes and delivers what they claim. We’ve been burnt repeatedly in the past on the big raises that weren’t tranched. These can’t be anorexic high frequency tranches of funding, but appropriately designed 9-18 month tranching is optimal.
- Later-stage ‘opportunistic’ or spec pharma deals are often much riskier than meets the eye. I’ve riffed on this theme before more generally (here), but we know this first hand because we’ve previously been seduced by their siren song: we’ve not done well with low-risk repositioned late stage assets like Ivrea, ARCA, Shogoo, Newron, Xytis, Nitec, etc from the 2003-2005 period. Our big takeaway lesson was to avoid these type of plays, and you won’t find them in any new deals since early 2007.
- Quick flips in biotech don’t happen often, unless it’s a flip to the trashcan. Promises of quick FDA approvals or near term M&A interest are most often empty claims. We’ve all heard the “this is the last financing before an exit” pitch, or “the ongoing Phase 2 is going well” in order to close a financing before opening the envelope. This latter pitch burnt us twice in the mid-2000s: Ivrea and Proprius both had inherited ongoing Phase 2s, both of which readout as completely uninterpretable due to protocol issues among other confounding variables. So much for the quick path to Phase 3. We’d prefer to focus on sustainable value creation as the thesis is much more credible.
- Company diversification by adding assets across risk or clinical stage often destroys value (another theme I’ve mentioned before). Forming a company with a Phase 2 in-licensed asset and a preclinical program isn’t smart due to the stage and value asymmetry: the Phase 2 program consumes the capital and dictates the future of the company, dragging the preclinical asset either up or down. Furthermore, even if the assets are of similar stage, they typically aren’t of similar merit. Adding a second asset to a story that’s not quite as compelling to the first just to make use of a management team feels like the wrong use of capital; figure out other ways of leveraging a team (e.g., part-time roles at other portfolio companies) and keep the bar for asset dilution very high. These observations 5+ years ago led to our asset-centric investment model around single program entities, like Stromedix’s STX-100 focus. Our AVDC initiative was born out of this observation in conjunction with Pharma’s externalization trend.
- Ultra-lean virtual models are great, but they have their limitations. For instance, big biology plays are hard to do without your own wet lab to do the bespoke work (though its fair to say we haven’t really gotten that wrong yet). Where we’ve felt some pain is around the constraints of the virtual model on other operational elements: when you have only a few key entrepreneurs in a startup, its very hard to go on the road for fundraising or BD without stalling or slowing the work on the R&D front. It can be incredibly distracting for a lean team to enter full due diligence with an army of experts from multiple pharmas or potential investors. Figuring out how to focus the BD campaign quickly is essential, and if one can avoid fundraising by having a strong insider syndicate that’s incredibly helpful. This is certainly something we are mindful of in our more virtual plays.
5. Lawyers are a necessary evil. We also looked at our legal, IP, and deal-making experiences to see if there were lessons here. Getting good legal advice is clearly very important to successfully navigating deals, especially in preventing bad outcomes from becoming worse.
- Patents are critical to biotech, so don’t skimp on doing deep IP diligence before investing. This has helped prevent us from doing some bad deals. We’ve only really had one big IP blow-up that changed a company’s trajectory: within months of the Dynogen $50M Series B closing in 2004 (no tranching!), a unexpected patent appeared that impaired the lead DDP200 program. So in some ways we’ve been lucky. But IP diligence goes beyond doing work before an investment; we’ve found that biotechs with best practice IP strategies often have an IP audit by an outside firm every couple years to ensure they are on strong ground
- For critical deals, investors should read the contract details. Similar to the diligence point above, trusting that others in the Boardroom have read the critical deal documents isn’t always a great strategy. Having a surprise appear after the fact can be a painful realization. We had this happen 5+ years ago a couple times with uncomfortable outcomes.
- Structured deals with Pharma have a long gestation time, no matter how many lawyers and accountants are in the room. Don’t assume you can speed them up too much. When Big Pharma XYZ tells you they are spinning out some assets and they show you the initial list, assume it’s going to take something in between a human and a pachyderm pregnancy to get to the closing, especially if there’s a structured buyback with complex consolidation details included. This gestation just comes with the territory of large bureaucratic organizations and complicated deal structures.
While we’ve had no major “Ah-ha” moments in reviewing our Lessons Learned (wish it were that easy), we do find that revisiting and updating them every few years is helpful. And during our champion and challenge process for new deals, we do bring up these lessons to gently bludgeon each other once in a while.