Prompt tracking is the foundation of every Rankry metric. Rankry holds a fixed set of buyer-intent prompts, runs them through each AI engine on a schedule, and structures the answers into data, so your visibility, position, and share of voice are all built from the same repeatable measurements.
For each tracked prompt and each model, Rankry records presence, position, sentiment, competitors named, and any cited sources, then stores the exact response text. Holding the prompts constant is what turns the inherent run-to-run variability of AI answers into a meaningful trend.
| Field | What it holds |
|---|---|
| Prompt | The exact buyer-intent question being tracked. |
| Category | The buyer-intent bucket the prompt belongs to. |
| Per-model result | Presence, position, and sentiment for each engine. |
| Competitors named | Rival brands that appeared in the answer. |
| Raw response | The full AI answer kept as evidence. |
Before a launch, a team adds the prompt "best [category] tool with API access" to tracking. Two weeks after publishing a developer page, that prompt flips from naming a competitor to naming them first on three of five engines, and the raw responses show why.