AI Readiness Audit / Method
Does an AI readiness score predict AI traffic? We checked.
Does a technical AI readiness score predict referral traffic from AI assistants? We compared audit results with real Google Analytics 4 traffic on five client sites. The question was not whether technical checks matter, but whether a higher score means more AI traffic once a site already passes the baseline.
What we compared
The set included an industrial valve manufacturer, a diabetes sports camp, a wealth-tech consultancy, a stone wholesaler and a simulation consultancy. All five had Google Analytics 4 connected. The window ran from July 18 to October 15, 2026.
Our traffic metric was sessions in GA4's "AI Assistant" channel as a share of organic search sessions. We compared it with the audit's technical score, citation share on twelve category prompts and question coverage, using Spearman rank correlation. With n = 5, these are directional findings, not statistical validation.
What the numbers said
Technical score versus AI share was about +0.4 on the first run and about -0.1 on the re-run with Search Console connected. Category-prompt citation share was -0.8 and -0.7. Question coverage was -0.1 and -0.7.
Same-day test-retest moved citation counts by 0 to 2 of 12. That variation did not explain away the negative sign. But a negative correlation in five sites does not mean citations cause less traffic.
The questions changed the picture
Every site already passed the technical baseline. Above that line, a higher technical score did not track AI traffic. Category questions measured upper-funnel presence, while AI referrals were driven by product and supplier questions.
Those are different measures of visibility. Being named when someone explores a category is not the same as being recommended when a buyer knows the product or supplier they need. A report needs to show which questions it asked before a citation count can be interpreted alongside referral traffic.
The wholesaler made the distinction clear. It was cited zero times on category prompts yet had the highest AI share: 5.3% of organic search sessions. Visitors landed on distributor and product pages because ChatGPT named it for product-level questions. Rewriting the prompts as a trade buyer sourcing wholesale moved it from 0 of 12 to 5 of 12, with no other change.
Why the traffic still matters
At the manufacturer, AI visitors converted at 5.6%, versus 1.4% site-wide. They generated 4% of all key events from under 1% of sessions. AI visitor engagement ranged from 60 to 90% across all five sites. Small volumes can still represent valuable buyers.
What changed in the audit
When Google is connected, prompts now come from Search Console question queries and AI landing pages. The audit detects brand mentions without links, and competitor suggestions account for business model. The technical score remains an actionable baseline with a published rubric, not a number to chase. Developers get the gaps; marketing gets evidence of who is named for buyer questions.
What this cannot prove
This is five sites, one 90-day window, three engines and twelve prompts. It cannot establish a general relationship or a causal one. The next check is at n = 8 to 10. Until then, read the results as a reason to ask better questions, not a promise of traffic.