01 · Discover
In-app search, category browse, and push deep-links. Watch query classes, zero results, and ranking-versus-inventory. A “healthy CTR” into thin listings is a false comfort.
Signal Map
Marketplace App Analytics, laid out as a map rather than a stack of KPIs.
This page is the studio’s public sketch of how we read two-sided apps. It is not software. It is the spine behind Listing to Loyalty and the shorter programmes.
02 · Evaluate
PDP, images, variants, reviews, and delivery promise. Marketplace App Analytics here asks whether bounce is UX, price, trust, or a seller who will not answer chat.
03 · Transact
Cart, payment mix, address friction, and cancellation immediately after pay. Separate wallet failures from promise failures. Thailand’s cash-and-wallet mix belongs in the model, not in a footnote.
04 · Fulfil
Slot accuracy, stock truth, returns, and seller cancellations. Ops metrics that never reach the product dashboard are still conversion metrics.
05 · Return
Second-order windows, category seasonality, promo hangover. Loyalty that ignores Songkran or payday is a smoothed fiction.
06 · Supply
Take-rate, featured placement, seller quality, and reply latency. Every buyer leak has a supply counterpart; Mirror Cloudcore refuses to teach one side alone.
How we use the map in the room
Cohorts pick one surface that hurts, walk adjacent surfaces only as far as needed, and leave with three leaks. The map stops people from “doing analytics” by adding charts until the original question disappears.
If you want to practise the whole loop, the flagship is Listing to Loyalty Studio. If only search is on fire, sit Search Ranking Diagnostics instead.
What the map deliberately omits
Paid acquisition auctions, brand-lift studies, and app-store keyword wars. Those can matter. They are not Marketplace App Analytics as we teach it, and mixing them into a first leak ranking usually hides the product problem.
See programmes Ask which surface to start with