Most AI-visibility dashboards lead with raw citation counts — the least useful number on the page. Across 284,425 citations we measured on four sites over 91 days, three KPIs actually predicted anything: share of authority (your percentage of all citations for your topic), the citation-to-session ratio (how many citations become real visits), and cited-page concentration (how few pages earn most of your citations). Track those three and you can steer; track volume alone and you are flying blind.
Every AEO and GEO tool now reports how many times AI engines cite you. That number feels like progress and tells you almost nothing. We know because we track AI citations across four of our own sites in different verticals, and over three months those sites logged 284,425 citations between them — enough to see which metrics move with reality and which just move. This is what the data says, not what a best-practices post guesses.
Why is citation volume a vanity metric?
Because it swings for reasons that have nothing to do with your performance, and because it hides where your citations actually come from. On the site we track most closely, weekly citations grew from about 2,000 to nearly 28,000 in eleven weeks — roughly 13.5× — and then kept climbing past 15,000 in a single day. Impressive. But raw volume alone would not have told us the two things that mattered: that the growth was real, and that almost all of it flowed through a tiny number of pages.
That concentration is the tell. On a peak day, 15,164 citations came from just 28 pages — over 500 citations per cited page. A handful of URLs carry the whole result. If you only watch the big number, you never learn which pages are doing the work, so you cannot do more of it. That is why volume is a vanity metric: it is real, but on its own it is not actionable.
What we measured across four sites
The 284,425 figure is not one lucky page. It is four sites in four different verticals — a B2B SEO/GEO SaaS, a legal-AI SaaS, an AI-tools site and a finance comparator — each tracked over the same 91 days. The spread matters: it shows the same playbook producing citations across unrelated markets, not a single fluke.
One site carried 154,834 of those citations on its own; the others logged 60,706, 47,670 and 21,215. Different sizes, same pattern: a small set of well-structured, answer-shaped pages doing most of the citing.
Which KPIs actually matter for AI search visibility?
Here are the metrics that survived contact with 284,425 real citations — what each one measures, why it matters, and how to track it.
| KPI | What it measures | Why it matters | How to track it |
|---|---|---|---|
| Share of authority | Your % of all citations for your topic | Competitive position, not just raw reach | Citation tracking, split by topic |
| Citation-to-session ratio | Citations vs actual referral visits | Exposes the no-click gap | Citations ÷ AI-referral sessions (GA4) |
| Cited-page concentration | How few pages earn most citations | Tells you where to double down | Citations ÷ distinct cited pages |
| Branded share of authority | Your % of citations on brand queries | Whether engines know who you are | Share of authority, filtered to brand |
| Crawler access | Whether AI bots can fetch you at all | The prerequisite for every other KPI | Server logs, or a free AI-readiness audit |
What is share of authority, and why does it beat raw citations?
Share of authority (SoA) is your slice of every citation in your topic — your citations divided by everyone’s. It is the difference between “we got cited a lot” and “we own this conversation.” On one site we track, SoA reached 19.9% of all citations in its niche; on brand queries specifically, it hit 37.9%. That second number is the one to covet: it means when people ask about that category by name, the engines hand the answer to us more than a third of the time.
Two competitors can both earn 10,000 citations a month. If one holds 30% share of authority and the other holds 4%, they are not in the same league — and raw volume would never show it. Share of authority is the KPI that turns a big, flattering number into a competitive verdict.
The no-click reality — and the ratio that reveals it
Here is the uncomfortable KPI. On one comparison site we track, roughly 6,600 citations in a recent 30-day window turned into well under 1% referral clicks. The engine answered the question using your content and the user never left the chat. That is the no-click paradox, measured — and it is why the citation-to-session ratio belongs on your dashboard. If you only celebrate citations, you miss that the value is shifting from the click to presence in the answer itself. You optimize differently once you accept that.
How do you track AI search visibility?
Two layers, and you need both. Track the humans AI sends you as referral traffic in GA4, and track the crawlers that read you in your server logs — we walk through both in how to track ChatGPT and AI referral traffic. And before any of it, confirm engines can actually reach and read you.
Every KPI above assumes AI engines can fetch your store. Confirm that first with our free AI Visibility Audit — it scores crawler access and server-rendered structured data in about a minute, so you know there is something to measure before you build a dashboard.
These 284,425 citations did not come from a secret. They came from one repeatable approach — server-rendered schema, a clear brand entity, and answer-shaped content engines can lift on the first fetch. That is exactly what the AI-Ready Kits build into a Shopify store, and what our done-for-you Expert service installs end to end. The verticals differ; the mechanics are the same.
A note on method. These figures are first-party: citations we measured on four sites we track, over 91 days ending 10 August 2026, using our own citation tracking. We report them anonymized by vertical rather than by name. One caveat in full view: a late-July to early-August measurement gap left several days uncounted on two of the sites, so we exclude those days rather than present them as a drop — the trajectory shown uses only complete-tracking weeks. Citation and share-of-authority figures reflect the tracked query sets and periods; they are not a guarantee of any specific result for another site. AI engines change how they surface and attribute content, so treat these as a measured snapshot, not a forecast. Last reviewed August 2026.
Across 284,425 measured citations, three KPIs actually steer: share of authority, the citation-to-session ratio, and cited-page concentration. Volume alone is a vanity metric. Confirm engines can read you with the free AI Visibility Audit, then build the schema, entity and answer content that earn share of authority with the AI-Ready Kits.