For agencies

Answer the client question you cannot currently answer.

Clients are already asking what ChatGPT says about them. Guessing is not a good look, and neither is a number you cannot show the working for.

What you get per client

Each client is a separate workspace with its own subjects, schedule and reports.

  • Recurring checks

    The same questions, asked on a schedule, so movement is measurable rather than anecdotal. A single reading cannot tell you whether your work did anything.

  • Reports with evidence attached

    Every number in a report traces to a stored answer. When a client asks where a figure came from, you can show them the text.

  • An outreach list, ranked

    The sources the engines already trust for that client’s service and area, ordered by how often they appear. That is a work plan, not a dashboard.

The outreach list withholds what it cannot prove

This is the part we would rather explain than have you discover.

  • Only sources proven to serve the UK are offered

    Cited sources for a UK question are not automatically UK businesses — we have measured an Alabama firm appearing in the sources for a query about a UK city. A target your client cannot realistically appear alongside is not a target, so any host we have not positively established serves the UK is held back.

  • And we count what we held back

    Withheld sources are shown as a count on the page — “N more cited sources held back until we can confirm they serve the UK” — rather than disappearing. A shorter list with an honest footnote beats a longer one with foreign firms hidden inside it.

Worked example

What monitoring one business for six days showed

A roadside-assistance firm in London, checked 28 times on the same three questions across six days in August 2026. Nothing was done to the business in between — no listings, no site changes, no campaign.

  • 17 → 50Range of the visibility score across those 28 runsOur corpus, August 2026. Mean 34.9. The business was unchanged throughout.
  • 28Readings behind that rangeSame three questions, same four engines, six consecutive days.
  • ±47%The swing either side of the mean a single check could have shownA check on the right day shows a 47% gain; on the wrong day, a 47% loss. Same business, no work done.

We are publishing this against our own interest: it means a single check — including ours — cannot tell a client whether they are improving. If an agency shows a score that rose after their work, ask how many readings sit behind each end of the comparison. Two points on a signal this volatile mean nothing. Read the full Q3 report.

Built for being questioned

Agency work gets audited by the client. These are the parts that matter when it is.

  • Every claim is traceable

    Answers and citations are stored and readable, so a report is a record rather than an assertion.

  • Unmeasurable is labelled

    A check that could not be scored shows a dash and says so, instead of presenting a confident zero you would have to defend.

  • No guarantees, on purpose

    We will never hand you a promise of placement to pass on. What you can promise a client is measurement, and that we do guarantee.

Try it on one client first

Run a free check on a client’s business, read the evidence, and decide whether the report is something you would put your name on.