Approach
A product is discontinued in March. The website says so. The feed still lists it, a retailer page still sells it, and in September an AI assistant still recommends it. Every one of those systems has an owner. The contradiction between them does not.
What we connect
What your organisation believes is authoritative about a product: the price, the specification, the availability, the claim you are willing to stand behind.
Your pages, your structured data, your feeds, your retailers, your syndication partners — each with its own update path and its own lag.
robots.txt, bot categories, CDN and WAF configuration, and whatever your platform added when it last updated.
Which identified services actually requested which pages, when, and whether the request was allowed.
How your brand and your competitors are described across a governed prompt set, and which sources are cited.
A prioritised backlog with owners and costs, maintained as the underlying systems change.
How we work
A substantial part of an early engagement is configuring the CDN, crawler controls, feeds, structured data, analytics and GEO tooling you already pay for, then testing what they actually do. We build software only where repeated engagements expose a missing join or a workflow that exists in none of the tools.
Configuration describes intent. What a crawler encounters is the outcome of several systems interacting, often including one nobody remembers enabling. We report what is observably true and treat the documented intent as a claim to be checked.
Edge telemetry shows that an identified service requested a page, when, and whether access was allowed. Prompt monitoring shows what AI systems say and cite. These are separate observations and we report them separately. We record what crossed, not what influenced.
Why it holds up
It would be easy to sell a dashboard that draws a line from a crawler request to a sale. The line would not be real, and the first customer to test it would find out.
A request proves a request. It does not prove that content was used for training, included in a particular answer, or caused a commercial outcome. Where we cannot see something, the report says so — unknowns are stated as unknowns rather than estimated into a number that looks complete.
A brand acting on our findings is changing feeds, crawler policy and product data across systems that are expensive to get wrong. Evidence that survives scrutiny is the thing being bought.
Standards
The telemetry is recorded againstContent Telemetry, an open standard governed by SPUR and published openly. We co-authored it and we operate one conformant implementation of it. We do not own it, our service is never a required destination, and a competing operator can implement the same standard on identical terms.
A measurement service you cannot leave is not credible as a measurement service. Your data is exportable and the format is not ours.
A fixed-scope review of one brand, product category and set of priority markets. You get an evidence-based baseline and a prioritised plan with owners and costs — not a report that sits on a shelf.
