Senior Product Manager, Service Delivery

The Role

This is a broad Senior PM role on the quality, trust, and communication side of service delivery — setting the right expectations on both sides, steering Pros to deliver great work, resolving conflicting interests fairly, and making our AI‑powered support genuinely good. You’ll work on live, high‑scale systems with a mandate to make them better.

Service delivery is a big area with more than one PM in it. You’ll work alongside them; your center of gravity is the trust between both sides of the marketplace.

• It’s a real area of impact, not a single feature. You’ll shape strategy, policy, and the systems behind a whole slice of the post‑booking experience — not optimize one screen.
• It’s inherently two‑sided. You work for customers and Pros at once, and you’re often the one arbitrating between them. Pleasing one side at the other's expense is failure.
• AI is a core tool, not a side project. Support and outreach already run on AI. You’ll push how far that goes — and build the evals that prove it’s good before we trust it with more.

Most service failures aren’t bad work — they’re mismatched expectations. The grass grew a tier past what was booked; “deep clean” meant something different to each side; the yard didn’t match the photos. You’ll make sure what’s promised to the Pro matches what the customer actually expects, that those expectations are the right ones for the property and service, and that Pros are steered toward delivering great work.

A customer wants a date the Pro can’t commit to. The grass is long enough to need a different price tier than booked. The job wasn’t what the listing photos suggested. These aren’t bad actors — both sides are right from where they sit, and the marketplace has to make a call. You’ll build the policies and the (increasingly AI‑assisted) decision systems that resolve these fairly and consistently at scale — and make the judgment call yourself when there’s no clean answer.

Customer‑side AI already handles ~34% of support at roughly a penny per message, and Pro‑side support is still to be built. Expanding either is gated on one hard, ongoing, technical thing: can we prove the AI resolved a case rather than just closed it? You’ll own the eval systems — golden datasets, automated judges, regression detection, human‑review sampling — that define what “good” means for an open‑ended conversation, gate every change, and catch quality drift before a customer feels it.

The inbox is how 500K+ customers and 20K+ Pros coordinate across all three brands — and the record we lean on when something goes sideways. It also has to moderate: catching when a relationship is drifting off‑platform (disintermediation) or a conversation is heating into conflict. You’ll keep communication easy for the legitimate 99% while spotting the patterns that quietly cost us customers, Pros, and revenue.

• Expectations match on both sides
  :
  Pros know exactly what each job requires, customers get what they expected, and fewer jobs go sideways from misalignment in the first place.
• Competing interests resolve faster and more fairly
  : A clearer resolution model with measurable consistency — and more of these conflicts prevented up front by better expectation‑setting.
• AI is provably good, and does more
  :
  An eval suite gates AI changes so no quality regression ships unseen — and on that foundation, higher resolution on customer support plus a first version of Pro‑side AI support live.
• Messaging connects and protects
  :
  Measurable improvement in messaging reliability and engagement, plus working moderation that catches off‑platform leakage and conflict early.

AI‑native. You use AI daily in your own work, and you have real intuition for how to measure whether an AI experience is actually good — not just whether it shipped. You’ve built or owned evals, or you’re hungry to, because you know that’s what…