agentos.guide › blog
AI visibility tracking means checking, on a regular schedule, whether AI engines name your brand, link to your pages and describe you accurately when someone asks about your category.
Book a free SEO strategy session →
Your Agent OS can carry most of that routine: drafting the prompt list, scoring the answers and writing the monthly comparison.
This guide explains the method first, then gives you three copy-paste prompts and a monthly loop. If you'd rather hand the whole thing over, book a free SEO strategy session with Goldie Agency.
Buyers increasingly ask an assistant before they search. If the assistant names three firms and you aren't one of them, the click never happens.
Google's AI features page says there are no special requirements for AI Overviews or AI Mode beyond sound SEO, so the underlying work is familiar.
What's new is the feedback loop. Without tracking you're guessing whether your pages, reviews and mentions are landing in AI answers at all.
AI replies don't come in fixed positions, so rank is the wrong unit. Presence is the right one.
Write these definitions into the agent's standing instructions so that every month is scored by identical rules.
Keep mentions and citations in separate columns. A brand that gets named but never linked usually needs stronger pages, whilst a page that gets linked without the brand being named usually needs a clearer brand presence on and off the site.
ai-visibility in your Agent OS workspace.customer-questions.md: paste in raw questions from emails, call notes and reviews. The rougher the better, because real wording is what buyers type.competitors.md: three to five rivals your prospects genuinely compare you with, plus the engines you care about.changes.md where you'll record every fix you make and the date.Book a free SEO strategy session →
That's the whole set-up. The prompts below do the heavy lifting.
🔥 Want these prompts pre-loaded and explained?
In the AI Profit Boardroom you learn AI SEO yourself, with the Agent OS ready to install, video walkthroughs of workflows like this one, and live coaching four times a week.
Read customer-questions.md and competitors.md in the ai-visibility folder.
Write 25 prompts a real buyer would type into ChatGPT or Google.
Make 17 unbranded (category, problem, local) and 8 branded
(reputation, reviews, alternatives, brand vs each competitor).
Give each an ID like CAT-01 or REP-03.
Use the customers' own wording where possible.
Save as prompt-set.md. Do not change it after I approve it.
Read the list before you approve it. Strike out anything a customer would never type, and swap in their phrasing where the agent has tidied it too much.
Put each prompt to each engine, or let the agent gather the replies if you've connected those tools, then save the raw text in a monthly file for scoring.
Read prompt-set.md and answers-2026-11.md.
For every answer, fill one row in the sheet:
mentioned = 1 if our brand name appears, else 0.
cited_url = any link to our domain, else blank.
competitors = every other brand named, separated by semicolons.
sentiment = positive, neutral or negative, with a one-line reason.
note = any fact about us that looks wrong.
Draft only. Do not edit the website or contact anyone.
Spot-check a handful of rows each month. Agents occasionally miss a brand tucked inside a comparison table.
Compare the November rows with the October rows in the sheet.
Read changes.md for what we fixed in between.
Report per engine: mention rate, citation rate and share of voice,
split by branded and unbranded prompts, with the sample size.
List only changes that also appeared in the previous comparison.
List every wrong fact in the note column and where it might come from.
End with three suggested fixes, ranked by effort. Draft only.
Book a free SEO strategy session →
The "also appeared last time" line is the important one. It stops you chasing noise.
📞 Turn the agent's report into a plan
Book a free SEO strategy session, share your first comparison, and we'll agree with you which three fixes deserve your time this month.
changes.md.The changes file matters more than it looks. Without it, next month you can't tell which fix moved which number.
Your prompt log is a sample. Two engines also publish first-party data about your own site, and the agent can fold both into the monthly report.
Google's Generative AI performance report counts how many times Google showed a link to you inside AI Overviews or AI Mode. You can slice it by URL, country, device or date, and Google says every website worldwide now has it.
Bing's AI Performance report counts citations across Copilot and Bing's AI summaries and samples the grounding queries behind them. Its Citation Share figure gives your percentage of the citations shown for one query.
Export both each month and drop the files into the folder. Neither one reaches ChatGPT, Perplexity or Claude, so the prompt log stays essential.
A prompt can return a different reply an hour later. Google explains that AI Overviews and AI Mode may fan a question out into several related searches, and may rely on different models, so their replies differ.
That's why prompt 3 only reports repeat changes. Run each prompt two or three times when you can, and judge trends across two or three months.
A zero on one engine is usually access, not quality. OpenAI's crawler page says opting out of OAI-SearchBot keeps a site out of ChatGPT's search answers, Perplexity says PerplexityBot is the crawler that surfaces sites, and Anthropic warns that blocking Claude-SearchBot may reduce visibility. Ask the agent to read your robots.txt against those three names.
Book a free SEO strategy session →
Begin with unbranded mention rate on each engine, because that's how new buyers find you.
Then check which of your URLs gained or lost citations. Pages that gain show the format engines trust for your topics, and they're the template for weaker pages.
Finally, work through the wrong-facts list and correct each item where it lives: your own site, a directory or a review platform.
If the sheet grows too big, a dedicated tracker can take over the sampling. Our best AI visibility tools guide ranks the options. For a one-off check of crawl access, entity facts and citations, run an AI visibility audit.
Want feedback on your definitions before month one? Bring them to a coaching call in the AI Profit Boardroom.
Give the agent the counting, scoring and comparing. Keep the prompt choices and the fixes for yourself or your team.
To learn the method properly, join the AI Profit Boardroom and learn AI SEO yourself. To have it run for you, book a free SEO strategy session.
📘 Free resources for your folder
Drop the free 200+ AI SEO Prompts into your workspace as extra prompt ideas, and read the free link building book on earning the outside mentions that lift share of voice.
Book a free SEO strategy session →
It can draft the prompts, score answers and write the monthly comparison. A person should still approve the prompt set and choose which fixes to make.
Around 25 is a good start, roughly two thirds unbranded. Keep the set fixed for a few months so the changes mean something.
Not with these prompts. Each ends with a draft-only instruction, so the agent reports and you decide.
No. It covers Google's AI Overviews and AI Mode only. ChatGPT, Perplexity and Claude still need your own prompt runs.
Prompt 3 only reports changes that repeat across two comparisons, because single runs vary too much to act on.
It records each fix and its date, so the next comparison can link a moved number to the work that moved it.
Set up the folder today, paste the three prompts, and claim a free strategy session whenever you'd like a second pair of eyes on your AI visibility tracking.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.
→ Best AI Visibility Tools In Your Agent OS: The Ranked List Plus 5 Copy-Paste Prompts
→ AI Visibility Audit: 5 Agent OS Prompts To Run It
→ Brand Visibility In AI Search: Track It In Your Agent OS
→ How To Get Cited By AI In Your Agent OS: A Quick Start, 5 Copy-Paste Prompts And A Weekly Loop
The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.
Join the AI Profit Boardroom →📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉