Meituan · Local Services · IM Consultation

Designing Trust Before the Bill

A 0-to-1 in-message quotation system across Meituan's 770M+ annual users and 14.5M merchants — turning uncertain local-service pricing into a guided, comparable, bookable decision. It's a super-app marketplace (Uber, Yelp and TaskRabbit in one) where deals close inside in-app chat, and I owned the whole journey: diagnose, compare, book, pay, review.

View Presentation Deck Try Prototype ↗

+30%

Intent→order conversion on the diagnostic channel — ~1.3× the old path. Routing price-anxious users into it lifted overall search conversion +0.5pp.

Product walkthrough

Context · Signal

The brief asked for price visibility. The evidence pointed deeper.

I needed a plumber. I messaged ten shops, actually talked to six, and spent half an hour comparing — and I still had no idea what it would cost. Every number felt like something they'd change once they showed up.

A pattern from user research — and my own experience

We shipped a standalone quote page first. Conversion didn't budge — the quote was rarely the final price, so users didn't believe it and merchants didn't maintain it. That failure was the insight: price wasn't a number problem, it was a process-trust problem — built in the conversation, not declared on a page.

Before · 4-step linear journey

01
Problem occurs
02
Many merchants appear
03
One-by-one outreach
04
Pick one for visit

After · 3-step trust loop

01
Diagnose the problem
02
Structure the intent
03
Compare and commit

System Design

One end-to-end flow. Trust compounds across every stage.

How a quote request becomes a booking, an on-site visit, and a settled order — across platform, user and merchant.

Open full diagram ↗

Quoting Engine

Conversation becomes a contract. Merchants quote against it.

Diagnosis start in chat — Vague problem → structured intent.

Post diagnosis recommendation state — Recommendations only after confidence.

Merchants quote the same diagnosis independently — so you compare guide prices against one order, never a platform-set final.

Show progress before price.

Live updates make waiting legible

Trust signals appear before price

A guide range, not a locked final

Reset only what is unsafe to assume.

Expired quotes stay visible but disabled

Only the time slot resets

Hard expiry, soft continuity

The trust loop closes where it began.

Return flow stays in the same thread

Re-engagement is one tap

The Other Side

Merchants quote against the same order — on equal footing.

Every merchant gets the same structured order and submits one quote — a fixed price or a tightly-bounded range, never open-ended. Since they can't see each other's numbers, they compete on the brief instead of undercutting — and for local repair, where the exact price isn't knowable until the visit, a bounded range is the honest unit.

Interactive Prototype

Try the full flow.

Switch scenarios from the rail, or tap the suggested replies to play a flow through. Re-skinned in English with USD placeholders; shipped in Chinese with RMB.

Open in new window ↗

Framework Extensions

The same loop scales: education, banquet, maternity care.

Home repair was the reference build. The same Diagnose → Structure → Commitloop maps cleanly onto other high-stakes, non-standard services — the substrate changes, the trust mechanics don't.

EducationBanquetMaternity care

Impact & Validation

Trust-first won the A/B.

+0%

Conversion lift · diagnostic channel

Measured

~60% of users ask about price before buying. Routed through a diagnose → quote → order flow, intent→order converted ~1.3× the old path (9%→11.7% toilet repair, 17%→22% pipe clearing) — lifting overall search entry +0.5pp.

~0k

Additional daily orders

Projected

−0%

Pricing disputes

Projected

A search-triggered floating window, piloted on two repair categories in Hangzhou and select Zhejiang cities. Conversion is real June–August A/B data; daily-orders and disputes are modeled forward for wider rollout.

Reflection

Next time, I would push on four fronts.

Next time I'd push on four fronts: give the merchant experience its own product pass; make guide pricing explain variability rather than imply a promise; scale with AI triage that escalates to human experts; and make cost ownership explicit when a diagnosis is wrong — who pays, who re-dispatches — so trust holds on the unhappy path, not just the happy one.

Transparent process is often a stronger trust advantage than transparent pricing alone.

Designing trust before the bill · 2025