# AI SEO

AI SEO creates a dated, source-backed baseline from supplied AI answers or a production-supported synchronous engine. It helps teams decide which pages, evidence, entity signals, and accurate third-party references are most worth improving.

## What it can deliver

- a bounded Query × Engine observation matrix with country, language, timestamp, mentions, citations, and failed or missing cells;
- citation-source patterns with explicit sample denominators;
- focused extractability findings for the brand's pages and pages cited in the sample;
- a prioritized GEO/AEO improvement plan with evidence strength, effort, owner, and a way to validate each action;
- an optional, separately labeled aggregate-mentions view when the supported DataForSEO dataset matches the requested market and platform.

## Example requests

> Test these three English questions for our brand in Google AI Mode for the United States. Show the exact quote before running the cells, then report mentions and citations without generalizing beyond the sample.

> I exported AI answers and citation URLs from Gemini and Google. Analyze them without buying new data and tell me which of our pages are hardest to extract or substantiate.

> Compare our brand with these two competitors in a small Google AI Mode sample for Germany in German. Keep AI visibility separate from general competitor positioning.

## Measurement and boundaries

The default paid baseline is up to three questions on one eligible synchronous engine. ChatGPT uses the DataForSEO consumer-answer scraper: production Search, request/response Schema and Quote verified on 2026-09-16; Use remains untested. Google AI Mode retains its successful production E2E record. Gemini has a verified scraper contract but execution is blocked until its quote discrepancy is explained or resolved and a fresh quote is checked. Claude live is excluded following the user's unavailable-path report; Perplexity has no verified synchronous path here. Supplied exports remain supported. Oxylabs LLM Realtime stays disabled. Each paid call needs a matching quote and covering authorization; reuse existing authorization and do not retry failures automatically.

Each scraper result is one provider-observed snapshot. A missing mention in one answer is not proof of general invisibility, and a citation is not an endorsement or stable rank. Country and language controls are reported as requested or returned. Optional LLM Mentions metrics query DataForSEO's pre-collected proprietary database: single-target aggregates provide a baseline; cross metrics compare 2–10 target sets. Their live label means synchronous retrieval, not a newly generated answer. Dataset counts stay separate from live observations and do not establish platform market share or provenance from customers' private request logs.

This skill is for AI-answer visibility and citations. Traditional technical/organic SEO belongs to SEO Audit, editorial roadmaps belong to Content Strategy, and general product/positioning comparisons belong to Competitor Profiling. The output is research and recommendations only: it does not publish pages, buy placements, contact publishers, or modify search, advertising, analytics, or CMS accounts.