, an innovative platform that connects healthcare providers and payers. But in my case, these risky investments involve 1’s and 0’s rather than poker chips and playing cards.
Fortunately at Cedar, we have a clear structure for managing risky ad hoc investments that guide us and helps our team scale up or divest from bets at the right moments—our Tech Bets ritual. Without it, we might find ourselves in a place plenty companies have been before, treating new opportunities like any regular “keep the lights on” investment. But bets have unique needs that require a structured and explicit process with custom procedures that are separate from, but complementary to, traditional planning.
Problem: Annual planning isn’t setup for surprises and ad hoc investments.
Wagers on new products and features are a part of nearly every company’s growth path, but they don’t always fit neatly into traditional planning timelines. At Cedar, we encountered this first hand in early 2023 as interest in Large Language Models (LLMs) surged everywhere, including healthcare.
launched, client executives inquired about how quickly we might be able to leverage this emerging technology to improve margins with automated service center calls. With our annual planning recently completed, it put us at an interesting juncture.
If it had been before we built out our Tech Bets ritual, we would’ve been in danger of falling victim to the same problems other companies face on a regular basis when making their bets:
Wrong framework: Without a specific alignment process, incorporating bets within your annual cadence is difficult, which can cause delays or a lack of investment in critical areas entirely.
Sunk-cost fallacy: When you make bets using a planning framework not designed for them, it can be hard to divest from bets quickly. Because without the right process and metrics, even if it’s fairly evident you should move on, a company may find itself continuing to invest in a project that should’ve been wound down.
Wrong support system: A bet may falter if nurtured like regular investments from the planning process. This doesn’t necessarily mean it was a bad bet, but rather they weren’t managed properly. I've been both the approver and proposer in these situations, and it's painful to see a project fail for this reason.
In the case of the sudden market shift with the release of ChatGPT, we had a strong incentive to move quickly and test our hypothesis: that a LLM based chat tool could create better margins.
But we couldn’t simply tack this on to our annual plan, because (1) it was already completed as mentioned and (2) we had additional considerations, like regulations and data sensitivity, that required fast validation and a clear procedure on whether to scale up, pivot, or divest from the bet entirely.
Solution: Tech Bets, a regimented planning structure for uncertainty.
Bets are inherently risky, fragile endeavors and it usually means the bigger the bet, the fuzzier the outcome. We’ve found you truly need structure and clarity to guide you through the process. You wouldn’t jump out of an airplane without a plan and a parachute, right? Same here.
Enter the Tech Bets ritual, a process that consists of
: pitching the bet, managing the bet, and pivoting or persevering.
While the top level steps feel fairly intuitive, the detailed structure is what gives the framework life. Each piece acts as another foot forward, bringing you closer to clarity on your bet. I walk through the entire process in even greater detail with my Elevate talk (watch the
: Define roles for decision-makers, stakeholders, content authors, and participants. Ensure all parties are informed of their role with appropriate notice.
: For each PPR meeting, ensure concise content is presented, where you consider the opportunity, the strategy (sometimes validated by sprints of research) and next steps - including whether to ultimately fund the bet.
Make sure the bet is reflected in Objectives and Key Results (OKRs): This is pivotal for accountability.
2. Manage the bet.
The same group established during the pitch process should be responsible for managing and nurturing the bet, being very intentional when you evaluate. The
Bets that have clear (tracked) milestones: Regardless of horizon, ensure you have a clear roadmap with milestone tracking (e.g., v1, v2, v3). Adjust based on progress and resource needs.
3. Pivot or persevere.
After approval is when the real work starts as you have to nurture the bet with the right kind of support. And this is where the more detailed procedures start to really shine as they guide your decision making. At your judgement points, you should discuss everything within your PPR and decide on the bet, whether the
It takes commitment, but once you’ve implemented the process, the results of the framework are very satisfying, even in the instance you kill a bet, because you’ve still gained clarity.
In our chat bot case, the first step meant defining things before deciding to approve. We created a detailed writeup (see
of the bet, the scope of the bet, the risks associated (see: regulations and data sensitivity) and the opportunities.
And once we approved it as a
@Horizon 2
bet, we made sure the investment was reflected back into our OKRs. I’ve mentioned this aspect a few times now, but we find it’s incredibly valuable to evaluate a bet based on it’s riskiness. Lower risk and less upside? More guardrails. Higher risk and more upside? Fewer guardrails (though still requires decision criteria).
After a period of time testing, the data came back on our chat bot showing little to no improvement in margins, and we were closing in on a key pivot or persevere moment.
, which are key as you work to isolate the results of the bet from other initiatives and more accurately measure its success (or failure). Persevering meant we pushed back our timeline by a month, but it ended up being worth it.
This was a critical moment because had we pivoted or killed the bet based on the initial data, we’d not only have a dead bet, but inconclusive answers about our original hypothesis.
This speaks to the importance of conviction in the face of ambiguity of your bets. Because even as you work to gain clarity, your faith in your bets may be challenged along the way.
In our case, the nurturing we gave our bet resulted in a successful validation of the hypothesis, albeit with a one month delay. But a slow good answer is better than a wrong fast one.
—a collection of actionable insights powered by the experience of the world’s best engineering leaders. Sign up for the Plato Community newsletter, powered by Coda, to receive more thoughts directly from Directors, VPs and CTOs on how they’ve improved their eng teams.
The goal is a process that works to bring structure and clarity to your bets.
What’s the best team size for the Tech Bets ritual?
Keep teams small, but not too small. A team working on bets shouldn’t creep into the 15-20 person range. Three to five is much more of the sweet spot. Less and you may in fact be under-resourced.
What should you do if you have to pivot?
Be explicit about and record your pivots. You don’t want the dedicated teams making changes without the wider teams knowing as it can start to feel like chaos, even if the right decisions are being made.
A big thank you to the following contributors on this post: Jon Collette, Aaron Zollman, Annie Lux, Craig Beveridge, Latham Arneson, and Justin Hales.