For finance teams
Finance can judge AI search visibility spend the way it judges any other programme: agree what will be measured and what counts as a change before the money moves, then read the result with its uncertainty. A monthly retainer should state its scope in writing, and a falling search budget should be rebalanced in stages, not in one move. Below are three answers for finance, then a plain guide to what each step of our own ladder buys and how to judge it after three months.
Judge it in three layers, and agree the tests before the spend starts.
The first layer is delivery. Did the agreed work happen? Ask for a dated log of every change: the page published, the listing corrected, the access setting fixed, each tied to the gap it was meant to close. Spend without a log cannot be judged at all.
The second layer is the answers themselves. A baseline is taken on a frozen set of buyer questions, and the same questions are asked again later of the same assistants. The result is reported as counts over their bases, for example named in 12 of 40 answers, with an interval, and each assistant on its own line. Finance should expect the interval, because answers vary from run to run and a change only counts when the two intervals do not overlap substantially. Flat results should be reported as plainly as the ones that moved.
The third layer is business signals: inbound enquiries that say they came through an assistant, and sales conversations where buyers mention one. Treat these as a floor, not as attribution. Many buyers who are influenced inside an assistant never leave a trace.
Do not accept a single blended score, one run on one day, or a revenue figure the data cannot support.
Read more: Reading an AI visibility figure honestly
A monthly retainer should say, in writing, what is measured and what is done. On measurement, expect an agreed set of buyer questions, asked again every month on the same wording, on named assistants, with repeated runs, reported as counts with intervals for each assistant separately. Expect a report that names a small number of prioritised actions, and a regular review call. If work is included, expect it to be named: a fixed amount of pages, sources or technical fixes each month, logged with dates.
A measurement-only step and a step with work done for you are different purchases. Check which one you are being quoted, because in the first the work that follows stays with your own team.
Before signing, check these things. What exactly is measured, on which assistants and how many times? Who does the work that follows? Is the scope, the provider coverage and the delivery timing agreed up front? What is the minimum term, and how do you leave? Do you receive the raw answers, or only a summary? How are changes made by the assistants themselves handled? Will flat results be reported?
Be wary of any retainer that promises a date or a number of recommendations. Nobody outside the assistant makers can promise what an assistant will say, and an honest provider promises the work and the measurement instead.
Read more: AI visibility measurement and the marketing work for B2B technology brands
Treat it as a reallocation with a name, not as an addition hidden inside another line. Give AI search visibility its own line with its own measures, and fund it by moving money from the programmes that are performing least, so the total does not simply grow.
First check the cause. A fall in search traffic has several possible explanations, and assistants are only one. Look at your own data by page type and by query type before moving money, because a decision based on a wrong diagnosis is expensive in either direction.
Within the new line, separate three costs. Measurement is small and follows the size of the question set. Work on your own evidence, such as specific pages and corrected facts, often draws on the same content and web effort that already serves search, because assistants that search the web read public pages. Earned sources, such as partner listings and independent write-ups, take time more than money.
Release the money in stages: a baseline first, then work against the gaps it shows, then a review on the same questions after a quarter. Keep the old channel's signals beside the new line so finance can see the shift without claiming it was caused.
Do not cut search fundamentals because assistants exist. Do not assume the traffic will come back either.
Read more: The first ninety days of AI visibility work
02 / Snapshot
30 to 50 buyer questions across three AI assistants, set against three competitors, with a standard report and a 30-minute walkthrough. The fee is credited against an Audit booked within 60 days.
£995
Book a Snapshot03 / Audit
100 to 250 buyer questions, observed answers, citation and recommendation analysis, a map of the evidence gaps and a prioritised 90-day plan.
From £4,500
Start with an audit04 / Monitor
50 buyer questions tracked every month, a monthly report with the five actions that matter most, and a call each quarter.
£595 a month, minimum 3 months
Start monitoringCreate your account in under a minute. Start with your first project and the questions that matter.
