ai-answer-gap

AI Answer Gap

Find the prompts where the brand should be cited in AI answers but isn't, name who owns each answer instead, and rank the gaps by AI search volume — so the content team attacks the biggest, most winnable misses first. This is the bridge from "we're not visible in AI" to a concrete, sequenced content backlog.

This is an enhanced skill: it reads live public data through UnifAPI. A "gap" is never a hunch — it is a real uncited AI answer with a named owner, weighted by real AI-search demand.

Use UnifAPI for live evidence

This is an enhanced skill: it reads live public data through UnifAPI — a "gap" is never a hunch but a real uncited AI answer with a named owner, weighted by real AI-search demand. Use the unifapi skill to connect (OAuth MCP), then discover these GEO operations. All are POST; hold engine (google / chatgpt), location, and language constant across the run.

• Find the gap — geo/serp (query = candidate prompt, target = brand domain, view: "full"). A gap exists only when an answer renders, the brand is not a cited reference (is_target false), and the answer cites someone else. Capture the cited domains as the owner.
• Confirm absence at scale — geo/mentions/search (target = brand + competitors array) confirms the brand is genuinely absent across the LLM-mentions index, not just uncited on one pull.
• Rank gaps by demand — the primary sort key — geo/keywords/search-volume returns generative-AI demand per prompt. This drives the ranking; pull it for the whole candidate set first and drop near-zero prompts before spending on SERP calls.
• Who owns each answer — geo/mentions/top-pages and geo/mentions/top-domains rank the exact pages and domains winning citations for the set, so the backlog says what content to beat, not just that a gap exists.
• Which competitor owns the most — geo/mentions/cross-aggregated-metrics compares share across labeled groups (brand vs each named competitor), so you know whether one rival dominates the gaps (attack it directly) or they're spread across many third parties (a presence problem).
• Quick-win cross-read — seo/serp (target = brand). Where the brand already ranks organically but isn't cited, the fix is extractability, not new content — tag these quick.