Marketing team / in-house marketer

AI search visibility for marketing teams

How do I brief this, how does it fit the SEO and content plan, and how will I know it worked?

Measured: at least 10 runs per engine, spent adaptively, across 5 AI engines · 95% confidence interval on every figure · open data (DOI 10.5281/zenodo.21612952 (opens in a new tab)) · how we measure

Take the brief template and the metric definitions.
Brief it as a measurement project with a content deliverable, not as a new channel. The measurable unit is share of model -- the share of AI answers naming you across a fixed prompt set run repeatedly per engine, with an interval and a capture date -- plus citation share, the share of cited URLs that are yours. Baseline it before anything ships, keep the prompt set frozen, and re-measure on the same set afterwards. That is the whole reporting design: without a frozen prompt set and a dated baseline, a later number is not comparable to an earlier one and the report cannot support the claim it will be asked to support.

How it fits the plan you already have

Most of the work is work your content and SEO plan already contains, aimed slightly differently. Pages that answer a buyer's real question in the first screen do better with assistants for the same reason they do better with people. The additions are narrow: keep a frozen list of the questions your buyers actually ask, publish something on third-party surfaces your category reads, and measure per engine instead of per keyword. Treating this as a separate channel with its own team and its own budget is the most common way it fails.

The brief you can hand to a contractor

State the category and the buyer, list the frozen prompt set, name the pages that should be able to answer each prompt, and specify the acceptance test as a re-measurement on the same prompt set at a stated date. Require intervals on every reported figure and a capture date on every number. Forbid guaranteed-placement language in the deliverable, because it cannot be honoured and it will end up in a client-facing slide. Ask for the raw runs, not just the summary -- a supplier who will not hand over the runs is asking you to take the summary on faith.

The metrics, defined

Share of model: proportion of answers naming the brand, per engine, across repeated runs. Citation share: proportion of cited URLs on the brand's domain. Answer presence: whether the brand appears at all, which is the only metric worth watching for a brand starting from zero. None of these is a rank, and none maps cleanly onto sessions, because an assistant answer often produces no click at all. Report them beside your existing organic metrics rather than inside them, or the two will be averaged into a number that means nothing.

Reporting it without overclaiming

Report the interval next to the point estimate every time, including in the summary slide, and date every figure at capture. Say explicitly when a movement is inside the noise band -- most one-month movements are. Attribute causally only where the change and the measurement were separated by a frozen prompt set and nothing else moved, which is rarer than it sounds. A report that says 'no separable change yet, here is the data' survives the quarter it is questioned in; one that claims a win it cannot evidence does not.

What counts as proof for this reader

The metric definitions and a brief they can hand to an agency or a contractor.

No figure is quoted on this page. The measured figures live where their runs, intervals and capture dates live: the AI Visibility Index and the methodology.

Take the brief template and the metric definitions.

A free AI-Visibility Teardown measures where you are recommended and where you are not — across the 5 AI engines — ChatGPT, Perplexity, Claude, Gemini, and Grok — plus Google's AI surfaces (AI Overviews & AI Mode), measured separately — reproducibly, with the raw runs attached. No call.

Start the free teardown

Reading for someone else? Find your fit.