TaskRabbit PM case: preventing and recovering from the most painful bookings

You are a product manager for TaskRabbit, the marketplace where people book local helpers ("Taskers") for jobs such as furniture assembly, mounting, moving help and cleaning. Work out which bookings go most badly wrong, and design product changes that prevent those experiences or repair them. Explain how you would choose which bad experiences to tackle first and how you would know your changes worked. ```hint Follow one booking end to end Trace a single job from describing the task to paying and reviewing, for both the client and the Tasker, and note every point where it can fail. Then weigh how often each failure happens against how much harm it does. ``` ### Clarifying Questions - Whose booking experience is in scope: the client who books, the Tasker who takes the job, or both? - Does "worst" mean bookings that never happen as planned (no-shows, last-minute cancellations) or jobs that happen but go badly (time and cost overruns, poor work, damage)? - Do we have data on where bad bookings cluster, for example by task category, city, or new versus repeat clients and Taskers? - Which outcome should the solutions serve first: client repeat bookings, overall trust in the marketplace, or keeping good Taskers on the platform? ### What a Strong Answer Covers - A definition of a "worst" booking grounded in TaskRabbit's in-home, two-sided marketplace, and a goal tied to it. - The booking journey mapped for clients and Taskers, with the specific points where it breaks. - A reasoned choice of which failures to tackle, based on frequency, severity and how much the product can change them. - Prioritized solutions that both prevent failures and recover from them, with their effect on Taskers considered. - Success metrics and guardrails, plus the risks the proposed changes introduce. ### Follow-up Questions - Your fix for Tasker no-shows is a stricter cancellation penalty. How do you keep it from driving Taskers away in cities where supply is already thin? - Most unhappy clients never leave a review. How would you find the worst bookings at scale? - One idea requires clients to upload photos of the job before booking. How would you test whether better scoping is worth the extra friction?