AI Visibility Results

What AI referral traffic actually looks like, with the real numbers

Most AI visibility claims arrive without evidence. This page shows the raw ChatGPT, Claude, Perplexity and Copilot referral data from my own portfolio, labelled by sector, for 28 May to 26 August 2026. The published figures are historical attributed visits, not current citation counts. Two of the sites have their own write-ups, with monthly data and limits: the EV charging directory case study and the health affiliate site case study.

Historical GA4 extractOwn portfolio, not client claimsVisits measured separately from citations28 May to 26 August 2026
See the AI Visibility Audit£1,500 fixed fee · delivered in 2 weeks

What this actually looks like, on real sites

These are 6 of my own portfolio sites, labelled by sector rather than name since most are not client work. Figures are the historical GA4 assistant-source session counts published for the 28 May to 26 August 2026 extract. They are attributed visits, not citation counts or proof of a ranking mechanism. Source matching can miss referrers and does not authenticate each visit as human. Shares are calculated from these published totals and rounded to one decimal place. See the referral study and its historical-window limitations.

1,357

AI-attributed sessions / May-Aug 2026 · Health & wellness

34.4% of the 3,948 published assistant-attributed sessions in this six-site historical sample, the largest count in the group.

993

AI-attributed sessions / May-Aug 2026 · Education & local services

25.2% of the published assistant-attributed sessions in this six-site historical sample, the second-largest count in the group.

593

AI-attributed sessions / May-Aug 2026 · Templates & productivity

15.0% of the published assistant-attributed sessions in this six-site historical sample came from this templates and productivity site.

570

AI-attributed sessions / May-Aug 2026 · EV charging directory

14.4% of the published assistant-attributed sessions in this six-site historical sample came from the EV charging directory.

259

AI-attributed sessions / May-Aug 2026 · Property investment tools

6.6% of the published assistant-attributed sessions in this six-site historical sample came from the property investment tools site.

176

AI-attributed sessions / May-Aug 2026 · Utility checker tool

4.5% of the published assistant-attributed sessions in this six-site historical sample came from the utility checker.

Why referral visits, search clicks and citations need separate measurements

A GA4 assistant-source session records a visit attributed to a matching source label. Search Console records Google search activity. A citation test records whether a particular answer links to a URL. A site can show different results across those measurements because they observe different events. These totals do not establish a source-selection mechanism.

The six-site extract is a historical sample from my owned portfolio. It does not represent all websites or forecast results for a client. Source matching can miss lost referrers and cannot verify that each session came from a person. The linked referral study provides the wider cohort, dates and limitations for its later edition.

Full methodology, including how repeated-run variance is measured for client engagements, is on the AI Visibility Audit page.

Frequently asked questions

Whose sites are these results from?

My own portfolio, not client work. Most are sites I own and operate myself, which is why they are labelled by sector rather than by name, and it is also why I can show the raw data honestly instead of a client-approved summary.

Does AI referral traffic prove that a page is cited?

No. These are recorded sessions with assistant-matched source labels. They do not count citations or identify why an assistant chose a page. Google clicks, assistant-attributed visits and observed citations need separate measurements.

How current is this data?

The six-site figures are the published historical extract for 28 May to 26 August 2026, 91 inclusive days. They are not a live dashboard. The linked referral study contains a later, different portfolio cohort and explains the overlapping windows and source-matching limitations.

Can you get results like this for my business?

Results vary by niche, starting point and how much of the technical and entity groundwork is already in place. The AI Visibility Audit measures where your business currently stands and what is realistic to fix first.

Related

Get Started

Want This Measured for Your Business?

Tell me your company website and the market you compete in. I will come prepared with an initial view of how you currently appear to AI assistants.

  • Focus on your biggest search problem
  • Review the evidence you already have
  • Identify the most useful next step
  • Answer your priority SEO question
15+ years experienceFree 20-minute diagnosisNo contracts

Usually responds within a few hours

Currently accepting new clients

Tell me about your project

No obligation · Direct reply from Sunny · Reply within one working day

Prefer to talk? 07305 523333 · Hello@SunnyPatel.co.uk