The Marketplace Priority Stack
Pre-commit an explicit ranking of who you serve first, so thorny trade-offs resolve themselves.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 4
- Confidence
- 90%
When user interest and business interest collide, most teams argue the specific case and let whoever has the most political weight win. Lovinsky's alternative is to establish guiding principles ahead of time that rank the constituencies of the business, then let the ranking adjudicate the individual fight. At Thumbtack the stack was customers first, pros second, Thumbtack last — a counterintuitive ordering for a marketplace where supply is usually the scarce side.
Origin
Noam Lovinsky, describing the operating principles used at Thumbtack while he was Chief Product Officer.
Core principles
- 01Guiding principles exist to resolve ambiguous or thorny questions, not to describe values.
- 02Serve the side whose satisfaction generates what the other side actually wants.
- 03Putting the company last in the stack is easy to say and hard to actually do in action.
- 04Optimizing for the business in the specific case is usually local-maxima behavior.
How to run it
- 1
Name the constituencies
List every party whose interests can conflict in your product — demand side, supply side, the platform itself, and any third party (advertisers, partners).
- 2
Work out what each side actually wants from you
Interrogate the second-order need, not the stated one. Thumbtack's pros did not fundamentally want features from Thumbtack — they wanted more, higher-quality customers. That reframing is what makes the ordering derivable rather than arbitrary.
Pro tip Ask: 'if I serve constituency A brilliantly, does constituency B automatically get what it wants?' Whichever direction that implication runs is your ordering.
Watch out Do not confuse who is scarce with who comes first. Supply is scarce in most marketplaces, but at Thumbtack customers still ranked above pros.
- 3
Commit to the ranking publicly, company last
Write it down and state it in the order it will be applied — at Thumbtack: customers first, pros second, Thumbtack last. The company goes last on the theory that if the first two are served well the business benefits downstream.
Watch out Saying 'company last' is trivial; honouring it when a revenue lever is sitting right there is the entire test of the framework.
- 4
Adjudicate live decisions against the stack, checking for local maxima
When a proposal helps a lower-ranked party at the expense of a higher-ranked one, reject it and say why. Lovinsky's canonical example: raising prices helps the business but creates liquidity problems in the marketplace — that is optimizing locally rather than globally.
Pro tip Make the local-vs-global framing the standard vocabulary of the debate so people can self-police before the decision reaches you.
Watch out Marketplace damage from a business-first decision is often delayed and diffuse (liquidity decay), so the P&L will look like the decision worked long before the harm shows up.
In the wild
Thumbtack's marketplace depends on pro supply, so the naive instinct is to serve pros first. The team instead reasoned that what pros actually need from Thumbtack is more high-quality customers — so building an outstanding customer experience that attracts and correctly matches demand is the most direct way to serve supply, and the business benefits as a consequence of both.
→ A written priority order the leadership team could apply to conflicts like pricing, rather than re-litigating whose interest matters most in every decision.
Raising prices was available at Thumbtack as an obvious business-positive lever, but it caused liquidity issues in the marketplace — fewer transactions clearing, a weaker matching engine.
→ Classified as local optimization rather than global optimization, and rejected on the strength of the stated principles rather than on a case-by-case argument.
Common mistakes
Assuming customer-good and business-good always coincide
There is high overlap between them, but not perfect overlap. Teams that pretend the conflict never happens have no mechanism for the moment it does, and default to whoever argues loudest.
Serving the scarce side first by reflex
Because supply is usually the constraint, marketplaces instinctively build for supply. But if supply's real need is demand, building for demand serves both — building for supply directly can serve neither.
Writing principles that never say no
A principle that doesn't rule out a decision you were otherwise tempted to make is decoration. The value of 'Thumbtack last' is precisely that it forbids the profitable-looking move.
Is it for you?
Best for
Marketplace and multi-sided platform product leaders facing repeated fights between user experience, supply economics and monetization.
Not ideal for
Single-sided products with no structural conflict between constituencies, or businesses in genuine survival mode where the company genuinely must come first.
From the transcript
“we had principles about which sides of the marketplace we wanted to serve in in which order”
“it was customers first, pros second and then Thumbtack last.”
“we we always just felt that what the pros need from us is more customers.”
“raising prices because we think it's good for the business even though it causes liquidity issues in the marketplace might be a little, you know,…”
From the episode
The happiness and pain of product management
Noam Lovinsky (Grammarly, Facebook, YouTube, Thumbtack)