The Adjacent-Precedent De-Risk Pitch
Win leadership buy-in for a big AI bet by anchoring it to a past bet that already worked.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 88%
Getting approval for a six-month ML bet with an uncertain payoff is a persuasion problem, not a technical one. Nika's approach: find an adjacent product inside the company that was an AI-first bet and succeeded, argue your proposal is structurally the same crazy-then-obvious bet, then explicitly cap the downside with a rollback plan and a stated maximum negative impact so leadership is deciding on a bounded loss.
Origin
Marily Nika's own method for pitching big AI bets to leadership at Google and Meta.
Core principles
- 01Adjacent, already-successful AI products are the cheapest source of both inspiration and credibility.
- 02Leadership approves bounded downside, not unbounded upside.
- 03Trust compounds with tenure — the longer you're at a company, the bigger the bet you're allowed to make.
- 04In a culture that welcomes failure, the pitch gets easier; in one that doesn't, the rollback plan does the work.
How to run it
- 1
Find the adjacent AI-first precedent
Identify a product the company already launched that was AI-first and succeeded — ideally one that seemed crazy when it was proposed.
Pro tip The more the precedent looked absurd at proposal time, the more persuasive it is as an analogy.
- 2
Frame your bet as structurally the same
Explicitly say: this seemed crazy at the time, here's how it worked out, and what I'm proposing is very similar to that crazy thing.
- 3
Attach a rollback plan
State exactly what you will do if the bet fails and how you unwind it, so approving the bet is not approving a one-way door.
- 4
Name the maximum negative impact
Quantify the worst case out loud — the maximum damage in a negative scenario — and show it is small enough that the expected value is effectively free.
Pro tip Naming the downside yourself is what makes leadership stop searching for a hidden one.
Watch out If you cannot bound the downside credibly, shrink the bet until you can.
In the wild
When Nika wants leadership approval for a large AI bet, she pulls up a prior AI-first launch, points out it seemed crazy at the time, maps her proposal onto it, then proposes a rollback plan and states the bounded maximum negative impact.
→ The decision reframes from 'do we believe in this uncertain model?' to 'do we accept this small, capped downside for a proven-shape upside?'
Common mistakes
Pitching the upside without the rollback
ML bets are inherently uncertain and can take six months to a year to resolve; without a stated rollback and capped downside, leadership hears an open-ended commitment and declines.
Ignoring the maintenance ask
Initial buy-in is often the easy part — the harder ask is keeping expensive people tweaking the model afterwards. Budget for that conversation, don't let it be a surprise.
Is it for you?
Best for
PMs at established companies needing executive sign-off for a multi-month, uncertain ML investment.
Not ideal for
Startups with no prior AI launches to reference, or teams where the AI investment is the company's only bet.
From the transcript
“people should know that there is an excellent source of inspiration and something kind of do risk things which is adjacent prodcts”
“I propose a little C CL like hey if that doesn't work out here's the roll back plan here's kind of the Maximum Impact you…”
From the episode
AI and product management
Marily Nika (Meta, Google)