Measurement and honest reporting
Generated answers vary between identical runs. Most AI-visibility dashboards sample each prompt once, which is why they move so much and mean so little.
Short answer
There is no submission form and nothing you can file that puts you in an answer, so nobody can show you a rank. Google's Search Console reports impressions in its AI features and nothing else: no clicks, no position, no queries. What can be measured is a fixed set of prompts, sampled repeatedly on a schedule, logged out, on a fixed locale, with the variation reported rather than hidden. That is what we do.
One run of a prompt is noise
Ask the same assistant the same question twice and you will often get different businesses named. Sampling a prompt five times produces a margin of error wide enough to swallow any change you made, which means a dashboard built on single runs is reporting its own variance back to you as progress.
We sample each prompt repeatedly per period, logged out, on a fixed locale, with a frozen prompt set and a frozen methodology, and we report the uncertainty next to the number. When a figure moves less than the margin, we say it did not move.
What we report
- Appearance rate. How often you are named across the prompt set, with the sample size and the spread.
- Description accuracy. When you are named, what the assistant says about you, and whether it is correct. Wrong hours in an answer cost more than absence.
- The competitive set. Who gets named instead of you, and which sources those answers are drawn from. This is where the next quarter's work comes from.
- Mechanical status. Whether the prerequisites still hold. Sites regress: a redesign reintroduces client-side rendering, a CDN rule gets switched on, a redirect chain appears.
Being cited and being named are different
A study of 3,981 domain appearances across 115 prompts found 61.7 percent of citations produced no brand mention at all. ChatGPT cited a source 87 percent of the time and named a brand 20.7 percent of the time; Gemini ran close to the inverse. If you sell something, the mention is the outcome that matters, and a report that counts only citations will flatter you.
We track both, separately, and we tell you which one moved.
The free first-party data nobody uses
Bing Webmaster Tools has an AI Performance report showing Copilot citations and the grounding queries the model generated. It is the richest first-party AI data that currently exists, it is free, and most businesses have never opened it. We set it up as part of the engagement whether or not you keep working with us.
Google Search Console's generative reporting is impressions only, with no clicks, CTR, position or queries. We report it for completeness and we do not pretend it is more than it is.
What we will not tell you
We will not give you a date by which the answers change. There is no index status to check and no platform publishes its retrieval cycle. We will give you dates for the implementation work, because that is ours to schedule. Treat any specific timeline promise, ours included, as a sales tactic.
Common questions
Because there is no ranking. A generated answer is composed per query, varies between runs, and differs by locale and by whether the user is logged in. The honest substitute is a frozen prompt set sampled repeatedly, which gives you a rate with a confidence interval rather than a position.
On a fixed monthly cadence, with the prompt set and methodology frozen so that periods are comparable. Changing the prompt set mid-engagement is the easiest way to manufacture an improvement, so when we do change it we report the old and new side by side.
Then we report that it did not move. The mechanical prerequisites are verifiable regardless, so you will always know whether the work was done correctly even in a period where the answers did not change.
What else an engagement covers
Want to know whether this is your problem?
Book a 30 minute call. We will fetch your site the way the crawlers do and tell you which of the prerequisites is failing. If none of them are, we will say so.