Case study: AI readiness of aisearchtune.com
The AIRA website goes through the same audit as our clients' sites. Below are real numbers from our engine's runs, unretouched.
Task
In August 2026 our own audit scored the site 53 out of 100 — the run of 2026-08-10 is published in the archived report. For a service that sells AI readiness, that is an unacceptable storefront, so we set the bar at 85+ under our own methodology — the live score is on the example report.
The latest showcase run scores 93 out of 100, and the report is open to anyone: the example report.
Approach
We followed our own report's recommendations without inventing anything: structured data (Organization, SoftwareApplication, WebSite, FAQPage, BreadcrumbList), the About and Who-it's-for pages, honest “suitable for / not suitable for” blocks, an objections FAQ, legal requisites and entity consistency across all pages, a complete PostalAddress and a service-area statement.
One standing rule: methodology false positives found along the way are fixed in the engine for every site — with pinned tests — never tuned to favour our own.
Result
53 → 65 after the first fix cycle (run of 2026-08-11), then 65 → 79 after the second — content per our own recommendations plus false-verdict fixes pinned on neutral examples (run of 2026-08-12). On 2026-08-17 the engine moved to methodology v2, and on the new ruler the same site scores 93 out of 100 (the latest showcase run). The remaining points are external signals: social profiles, reviews, inbound links. Every number is reproducible: these are real audits by our engine.
The score of 53 was measured with methodology v1 (August 2026). From version 2.0 the score is computed differently: AI criteria carry it, and the basics add nothing — so this case study’s numbers and the site’s score today come from different rulers. The current score is always visible in the example report.
Products used: Basic AI Readiness Scan and Full AI Recommendation Audit — the same plans available to clients.
Why we publish this
A service that tells others to be verifiable by AI must withstand its own verification. This case study is a working log, not a polished success story: the numbers update as the work continues.