Store Performance

Every product card and query row in Biasque is built from the same two primitives:

• Appearances (hitCount). The number of searches in which the AI agent considered this product. Appearance means the agent looked at the product as a candidate — selected or not.
• Selections (selectedCount). The number of searches in which the agent actually recommended this product to the user.

Everything else on the dashboard is derived from these two numbers.

selectionRate = selections / appearances, shown as a percentage. A selectionRate of 50% means that whenever the agent looked at this product, it picked it half the time.

Read it carefully:

• High selectionRate, low appearances — the agent likes you when it finds you, but it rarely finds you. Discoverability problem (title, description, category, price).
• High appearances, low selectionRate — the agent finds you a lot but keeps picking someone else. Positioning problem (you are visible but losing the comparison).
• Both high — you own the intent. Make sure inventory and pricing can keep up.

The dashboard shows categoryAvgSelectionRate next to your own number. This is the average selection rate across all products in the same category over the selected period.

Always compare to this baseline before celebrating or panicking. A 30% selection rate is great in a crowded category where the average is 12%, and mediocre in a niche category where the average is 55%.

For each product (yours and each competitor) the dashboard lists the queries that drove the most appearances. Each row shows:

• Query — the natural-language search the agent received.
• Appearances — how many times the product surfaced for that query.
• Selections — how many of those resulted in a recommendation.
• Selection rate — selections / appearances for that specific query.

Use the per-query breakdown to find your strongest intents (queries where you win) and your soft spots (queries you appear for but lose). Competitor query lists tell you exactly which intents you should be competing on but currently are not.

The trend chart plots appearances and selections over the selected window (7 / 30 / 90 days). Use it to:

• Confirm whether a change you made — to the product title, description, or price — moved appearances or selections.
• Spot competitor moves: a sudden drop in your selectionRate often coincides with a competitor improving their listing.
• Catch seasonal effects before they show up as "we're losing" in a flat number.

Where a merchant exposes an MCP endpoint, the dashboard also fetches livePrice, liveCurrency, and liveInStock at view time. PossibleliveDataStatusvalues:

• fresh — live data fetched successfully.
• no_mcp_endpoint — the store does not expose an MCP endpoint.
• no_variant_gid — we could not extract a variant identifier from the product's checkout URL.
• mcp_error — the MCP call failed (timeout or upstream error). Catalog price is shown instead.

Everything on the watchlist dashboard is also available over HTTP via the Monitor API.