L
Lustre
Product architecture

How Lustre works under the hood

A walkthrough of the core flows, services, and the matching engine that pairs customers with the right personal shopper.

Three personas, one marketplace

Customers

Post requests, browse the catalog, message shoppers, and approve orders.

Personal Shoppers

Claim briefs, source pieces, manage fulfillment, and earn service fees.

Admins

Operate the marketplace — users, catalog, moderation, analytics.

Core request → delivery flow

Customer posts request

Title, category, budget, notes

Matching engine

Scores shoppers by specialty, locale, rating, capacity

Shopper claims brief

SLA timer starts

Sourcing & curation

Lookbook, negotiation, in-app chat

Order placed

Escrow hold, invoice generated

Fulfillment

Confirmed → Shipped → Delivered

Matching engine — rules

Every open request is scored against the active shopper pool. The highest score wins first refusal, with a 30-minute SLA before re-routing.

30%

Category specialty match

Shopper's primary tags vs request category

25%

Locale & vendor access

Proximity to ateliers, language fit

20%

Past performance

Conversion rate, on-time delivery, CSAT

15%

Capacity

Open briefs vs declared bandwidth

10%

Rating recency

Weighted 90-day rolling rating

Platform services

Marketplace API

Requests, orders, listings, users

Matching service

Scoring + routing rules

Conversations

Real-time chat & threads

Payments & escrow

Holds funds until delivery

Notifications

Email, push, in-app

Catalog search

Faceted, relevance-tuned

Trust & safety

Moderation, fraud heuristics

Webhooks

Outbound events for partners

Lifecycle funnel

Request → delivery (last 90 days)