App Marketplaces Lab guide

Start with the sample

Open the sample: a Shopify inventory category and a WordPress booking category with generated apps, prices and reviews. It works offline and fills every view.

Watch a category and sweep it

Save a watch: a store, a category, how many reviews to read per app (25, 50 or 100) and how many list pages (1 to 4). A sweep has three steps: the category's app lists, the details of each app, then its 1 to 3 star reviews. Over 200 requests it asks you to confirm first. If a store slows requests down, a banner says so and the sweep goes on more slowly. A step that reads nothing for 4 minutes is stopped as a recorded stall, and you can resume it.

The four views

• Landscape: reviews (log scale) against rating, dot size for installs where the store reports them, fill for free or paid. A shaded Popular but poorly rated region prints both of its lines. Lasso a set of apps to list them.
• Pricing: a ladder with one row per app and a dot per paid tier on a shared monthly price axis. Free tiers sit in a Free column, with a summary of free only, free plus paid, paid only and not stated. Prices of paid apps are read on request.
• Pain: a grid of apps by theme with counts, a ranked theme bar, and word for word quotes under the chosen theme with the matched phrase marked. Pin a theme or an app.
• New: apps by launch date with reviews gained per month, plus what changed since the last sweep: new, dropped, rank moves, review growth and rating change. The first sweep is a baseline only.

Honest limits

• Shopify has no install counts. Review count stands in for demand there, and it is a proxy, not a user count.
• WordPress installs are bands such as 10,000+, shown as the band and never as an exact figure.
• Atlassian has no launch dates, so its apps cannot be placed on the launch timeline. Installs are shown where the store reports them.
• Where a store does not report a figure, the Lab says not reported. It never prints zero.
• The opening is a region, not a score. Popular means at or above the middle of the category on reviews, with a floor. Poorly rated means under a fixed rating line. Both lines are printed on the chart.
• Reviewer names are never stored. Pain themes count phrases, so a match is a sign, not proof.

Pin, watch and ask

Pin an app or a pain theme with its quotes to the evidence board. Turn on the weekly re-sweep for a watch (only while Desktop is open). Every action is an MCP tool: see the App Marketplaces Lab group in the tools reference. For one-off fetches of a store, see the Research sources page.