AI Visibility Audit

AI Visibility Audit reads how AI answers about your market mention and cite your brand, then recommends one to three changes. A change might be a third-party page to get onto, a page of your own to improve, or an access problem to fix. Each recommendation comes with the answers and cited pages behind it.

Use it when you want an AI visibility, GEO or AEO audit. To research what people ask before you track anything, use AI Prompt Research.

Before you start

• It needs the Business overview, your website and your main competitors in project context.
• Your tracker's own brand name is the project name. If that isn't how people write your brand, the assistant asks you to rename the project before it collects or reads answers.
• Saving needs a connection with Make changes on.

Where the answers come from

The audit reads the answers your prompt tracker collected from the engines it covers. Trackers made in the onboarding store setup (See how AI recommends your store) cover ChatGPT on the free plan, and ChatGPT, Gemini, Google AI Overviews and Perplexity on Pro. A tracker the assistant creates for the audit starts with ChatGPT and Gemini on Pro, or ChatGPT on the free plan, unless it picks other engines; Gemini, Google AI Overviews and Perplexity need Pro. ChatGPT and Gemini answers are captured from their apps via DataForSEO, Google AI Overviews answers come from Google search results via DataForSEO, and Perplexity answers are API answers via DataForSEO that can differ from the Perplexity app.

Prompt Explorer answers come from the model's API, not the consumer app. The audit uses them to explain why pages get cited and never counts them in mention or citation rates.

What it does

1. Gets evidence. What it does depends on your tracker:
   • A completed run: it uses that run and buys no new answers.
   • A run in progress: it follows that run with get_ai_visibility_run.
   • Active prompts but no runs: it estimates one check, asks you once, and runs it with the approved one-off cost as the limit.
   • No active prompts: it asks you before resuming anything.
   • No tracker: it proposes 10 to 15 neutral prompts across two or three topics and shows you the cost of one check and the monthly cost. After one approval, it saves the tracker and runs one check. If you want a one-off audit, it pauses the schedule. If you decline the spend, it audits on prompt research alone.
2. Reads the results for each topic and engine, and finds gap prompts: prompts where a competitor is named and your brand isn't.
3. Pulls the run's cited sources once, classifies the 10 to 15 that recur most, reads five or six of them and two or three gap answers, and runs Prompt Explorer on one or two leading gap prompts.
4. Checks your side. It reads your robots.txt for AI crawler rules (GPTBot, OAI-SearchBot, ChatGPT-User, Google-Extended, PerplexityBot, ClaudeBot), compares your pages with the pages cited instead, and runs live Google results for one or two of the fan-out queries Prompt Explorer returned. The assistant reads web pages with its own tools, outside NameIO.
5. Shortlists five to eight actions of at least two kinds, then chooses one to three.

What it costs

• With a completed run, the audit buys no new answers.
• A paid check runs only after you approve the estimate once. The approved one-off cost becomes the check's limit, and the check is refused if it would cost more. Only one check runs per project at a time.
• Saving a tracker uses no credits. A new tracker runs weekly, with its first scheduled check a week later. Scheduled checks need Pro: on the free plan a tracker runs only on demand, and the tracking page shows "Scheduled checks are part of the paid plan. Upgrade to run them on a schedule, or use Run now."
• On the free plan, a one-off check starts only if it fits in full in today's free data budget, or your organization's top-up credits cover it.
• Prompt research and Prompt Explorer need Pro. Before the first Prompt Explorer call, the assistant tells you it uses a small amount of credit.

What it saves

• Project context: the pages the report names, as key pages, and a research-log line.
• The report.
• When there is no tracker, it saves one (save_ai_visibility_tracker) and pauses its schedule (set_ai_visibility_schedule) if you want a one-off audit. When a check is needed, it runs one (run_ai_visibility_check), which spends credits.

The report

The report is titled like "AI Visibility Audit — Oct 2, 2026" and has five sections:

1. Where you stand in AI answers.
2. Recommendations: one to three, each with Do this and Why, and an evidence table.
3. What AI cites in your market.
4. What else we checked.
5. How this report was made: the run ID, date, market, engines and coverage, a link to this page, and an "Evidence and methodology" block.

If the SEO Report skill isn't available, the assistant says so and stops.

What it won't claim

• It doesn't invent visibility scores, share of voice or sentiment.
• It claims a change over time only from get_ai_visibility_trend, and only when that tool reports the runs as comparable.
• With fewer than about ten neutral prompts, it says the sample is small.

Tools it uses

get_ai_visibility_tracker, get_ai_visibility_results, get_ai_visibility_sources, get_ai_visibility_answer, get_ai_visibility_trend, research_ai_visibility_prompts, explore_prompt, estimate_ai_visibility_cost, save_ai_visibility_tracker, set_ai_visibility_schedule, run_ai_visibility_check, get_ai_visibility_run, get_ranked_keywords, get_serp_results, get_backlinks_overview, list_projects, get_project_context and update_project_context, plus save_report through SEO Report.

See AI visibility for how tracking and checks work in the app.

What's not done

NameIO doesn't offer a way to install this skill yet. See Skills.