This is a template, not the training
Every training I run is tailored to the client: your tool, your prompts, your markets and the people in the room. In the live session we read your own data, not these invented numbers.
Use this deck to see the approach. The live session, built on your tracking data, is a different thing.
06.
Measuring AI visibility
Five numbers from an AI visibility tool
Most AI visibility trackers report the same core metrics, under slightly different names. Use one set of terms throughout, so every report can be checked against the tool.
Based on presence and top positions across results. The more often and the higher the brand or domain appears, the closer to 100.
Share of prompt results in which the brand is mentioned. The main metric for recommendation-type questions.
Share of prompt results in which the brand's domain appears as a link. The main metric for educational questions.
Whether sources mention the brand in a positive, neutral or negative context. Ranges from −1 to +1.
Average position of the brand's URL across the engines analysed. Does not include the position of brand mentions.
IMPORTANT: in the first weeks after setting up a tool, read levels, not trends. Trends become meaningful after a few months of data.
Swipe sideways to see the whole table →
Cited as the expert, named as the answer
Not every question should end with the same win. Judge each prompt by the metric that matches its purpose — most tools can tag prompts by funnel stage.
“Which hand cream is best for cracked fingertips?”
0 — the brand's site is not a source
Brand A #1 in Google AI Overviews, #2 in ChatGPT
Success. For this question the recommendation is what counts, not the link.
GOAL: BRAND MENTION — be the recommended product
EDUCATIONAL QUESTION · ToFu
“Why does the skin on my hands crack in winter?”
the brand's dry-hands guide in AI Overviews (#4), homepage in AI Mode (#2)
Brand A named in AI Overviews
Success of another kind. AI learns from the brand's page — that builds expert status, but the answer doesn't end in a purchase.
GOAL: DOMAIN PRESENCE — be the source AI quotes
PRIORITY: purchase questions first — that's where the brand is chosen. Educational questions build the expert status that supports those recommendations.
Swipe sideways to see the whole table →
BRAND MENTIONS VS SOURCES · ILLUSTRATIVE
Recommendation leader, not yet the expert source
Among brands named by AI, Brand A leads clearly. Among websites AI cites, it is second — behind the national health portal and ahead of large medical sites.
BRAND MENTIONS · VISIBILITY SCORE
Who AI names in its answers
SOURCES & LINKS · VISIBILITY SCORE
Whose website AI cites — 640 domains in total
WHAT IT MEANS: the real rivals are not other brands but the generic ingredient and the health authorities. Keep the recommendations, and close the gap to the health portal as the source for educational questions.
Illustrative data, one market, one month. Brand mentions cover generative engines only.
GEO AUDIT · HOMEPAGE EXAMPLE
Step by step, page by page
Good tools audit each page for crawlability, structure and readability. The goal is not 100/100 everywhere — it's steady improvement on the pages that matter.
FIX IN THE CMS · LOCAL, PER PAGE
Done by the market team as part of content optimisation
• Alt text missing on images (38 on the homepage)
• Q&A blocks and tables — where they fit the page type; not every page needs them
• Key messages shown only in images move into page text
• Schema type chosen and added for each page
FIX ON THE PLATFORM · GLOBAL, ONCE
Raised with IT / the web agency, benefits every page
• Response time (1.3 s, flagged)
• Content signals in robots.txt (search, AI input, AI training) — a policy decision
• Markdown negotiation and link headers
• Hero banner rendered by JavaScript — invisible to AI crawlers; how images load
HOW TO USE IT: audit each priority page before and after, fix what the CMS allows right away, and bundle platform fixes into one request to IT.
Why AI says what it says about the brand
Some tools also show how AI describes a brand — what it praises, what it criticises and how it rates it against competitors.
PERCEIVED PROS BY CLUSTER
Positive themes grouped into clusters, compared across brands.
Illustrative layout — no data
PERCEIVED CONS BY CLUSTER
Negative themes grouped the same way — the source of a low sentiment score.
Illustrative layout — no data
Scores from −5 to +5 on the attributes that matter in the category, side by side with competitors.
Illustrative layout — no data
WHAT TO DO: set up attributes and competitors per market, turn every 'con' AI repeats into a content gap your guidelines answer, then track whether perception shifts.
One tool, every question answered
Beyond the headline numbers, each view in the tool answers a specific question — and feeds a specific step of the work.
How does each question perform?
→ Selecting the priority pages
Which of the brand's pages are already cited?
→ Finding what works and reusing it
Who does AI cite instead of us?
→ Knowing which sources the content must match
Who is named alongside the brand?
→ Tracking recommendations vs competitors
Where are competitors present and we aren't?
→ A ready list of gaps to close
Which sub-questions does AI search for?
→ Which sections and FAQs a page needs
PLUS: a page audit for each priority page and a brand perception report for how AI describes the brand — ideally in one tool.
What the data tells us to do
A first look at a few weeks of data — enough to see where to start.
Make the website the source
In 5 of 7 markets AI names the brand far more often than it cites the brand's sites. Educational pages need answers AI can quote.
Protect the recommendation
The brand leads mentions. The challengers are the generic ingredient and health authorities — product pages must say clearly why this brand.
Fix pages locally, the platform once
Most issues are fixed in the CMS, page by page. Platform issues go to IT in one request and help every market.
Measure with tools you already own
An AI visibility tracker for AI answers, Search Console and GA4 for traffic — same prompts, before and after. Brand perception next.
NEXT: turn these findings into a short list of priority pages per market — chosen from the prompts above.
Who this session is for
Brand, SEO, content and analytics teams that already track AI visibility in a tool, or are about to start. It works with most trackers on the market: the metric names differ, the logic is the same.
It is one module in my SEO & AI Search training. If you are choosing a tool or a method, see GEO and AI visibility.
