Building trust into a two-sided marketplace: lessons from 18 months of platform design

The central challenge of any pet care marketplace is trust. Pet parents are handing their keys to a stranger and leaving an animal that can't communicate distress. The stakes are higher than most gig economy services, and the trust infrastructure needs to reflect that.

Here's what we learned building TOOF's trust system over 18 months — what worked, what didn't, and what we're still figuring out.

Why we made meet-and-greets mandatory

In v1, meet-and-greets were optional. Conversion was higher — roughly 30% more bookings completed in the flow. But first-visit cancellation rates were also 3x higher, and post-first-visit rebooking was significantly lower.

The friction of the mandatory meet-and-greet wasn't killing conversions — it was filtering for the right bookings. Parents who completed a meet-and-greet and still booked had meaningfully higher lifetime value, lower cancellation rates, and dramatically better NPS scores.

We made them mandatory in month 6 and never looked back. The conversion dip recovered within 8 weeks as word-of-mouth quality improved.

Designing sitter profiles for the right signals

Our first sitter profile design emphasized credentials: background check badge, years of experience, number of pets owned, certifications. This made sense on paper.

But when we ran user research on what actually drove booking decisions, credentials were table stakes — not differentiators. What parents actually used to decide: specific species and temperament experience, the tone of their written bio, photo quality, and response time.

We redesigned profiles to lead with a narrative bio (required, minimum 150 words), species and behavior experience checkboxes that feed into matching, and a response time indicator based on actual data. Conversion on profile views to bookings increased 40%.

What we still don't have right

Reviews are still too soft. Most reviews cluster at 5 stars with low-signal text. We're experimenting with structured review prompts that force more specific feedback, but the UX tradeoff between completion rate and quality is genuinely hard.

We're also still working on how to surface "quiet but excellent" sitters — people who are consistently great but don't have the marketing savvy to self-promote their profiles. Algorithmic ranking based on behavioral data (rebooking rate, response time, recap quality) is part of the solution, but we're not fully there yet.