For CMOs and marketing leaders

AI visibility for CMOs and B2B marketing leaders

A B2B technology CMO should treat AI search visibility as a measured budget line with a small first commitment, sized from the number of buyer questions to be tracked rather than from a share of the media budget, and reviewed against evidence after a quarter. The risks are specific and manageable, the choice between an agency and an in-house team turns on who can do the evidence work, and the board needs counts with their limits, not a score. The answers below cover each in turn, followed by a one-page board case you can adapt.

Talk to us See all services

A pale flamingo standing on bright yellow legs.

How should a B2B technology CMO decide how much budget to put into AI search visibility next year?

Start from the job, not from a percentage of the marketing budget. We know of no reliable published benchmark for what share of a marketing budget AI visibility should take, so a number borrowed from another company's plan has nothing behind it. Size the job instead. Count the buyer questions you would want measured, multiply by the assistants you care about and by how many times each question is asked, and you have the size of the measurement. That is the part of the cost that follows the size of your question set, not the size of your company.

Then split the year into stages with a decision at the end of each. First, a baseline: where assistants name, cite or skip you today, on a frozen set of questions. Second, work against the gaps the baseline shows, which costs your own team's time or an agency's fee, and is a separate line from measurement. Third, a repeat measurement on the same questions, so the board can see what moved and what did not. Hold part of the year unallocated until the baseline is in, because the baseline decides where the work goes.

Do not tie the budget to a promised result. Nobody outside the assistant makers controls what an assistant recommends, so what you are funding is better evidence and an honest measurement of it.

What risks does AI-assisted vendor research create for a B2B software brand, and what should marketing leaders do about them?

The first risk is absence. Buyers put questions to assistants that name no company, such as which tools suit a team of their size, and the shortlist forms inside the answer. A brand that is not connected to the question in the evidence an assistant can find is simply not on it, and nobody tells you.

The second is being named but described badly: an out of date capability, a wrong sector, a rival's framing. The third is a rival whose public evidence is clearer or more independent than yours, so the answer draws on theirs. The fourth is variation. The same question can return different names on different days, so one check, or a check on a single assistant, misleads in both directions. The fifth is blindness: you cannot see what a particular buyer was shown.

What to do about them is mostly unglamorous. Agree the buyer questions and ask them repeatedly, on each assistant separately. Read the sources each answer cites. Fix your own evidence first: specific pages, the same facts stated the same way everywhere, and a website that the assistants' search crawlers are allowed to read. Earn independent sources by giving others accurate material. Then measure the same questions again.

None of this removes the risks, because the assistants are not yours to control. It makes them visible and shrinks the part that is your own doing.

Should a B2B technology company use an agency or build an in-house team to improve how AI assistants describe it?

It depends on who will do two different jobs, and most companies answer each differently. The first job is measurement: asking the same buyer questions of several assistants again and again, recording every answer and counting what comes back. The second is the work that follows: writing the specific pages, correcting facts across your listings, fixing access settings and earning independent sources.

An in-house team suits a company with a content owner, someone who can edit the website and change its access settings, and the patience to run a small question set by hand or with a tool. You keep the knowledge and the control. An agency suits a company that needs a baseline quickly, has no spare capacity in the team, wants the people who read the numbers to be the people who act on them, or wants an independent reading before committing.

Many companies split it. Some buy a baseline and a method from an agency, then take the work in-house. Others use a tool for the repeated asking and do the rest themselves.

We are an agency, so weigh this answer accordingly. Put the same two questions to every option, ours included: what exactly is measured, and who does the work that follows?

A one-page board case for AI visibility work

The question the page answers: where do buyers who ask AI assistants about our category see us, and what do we propose to do about it? Write that as the first line.
What to report: three counts for each assistant on its own, never added together. How often an answer names us, how often it puts us forward as the choice or a named alternative, and how often it cites our own pages. Write each as N of M with its interval beside it.
The trend: the same questions, the same assistants and the same wording, period after period, one line per assistant. If the questions change, start a new series and say that the old and new are not comparable.
What is uncertain: the interval, any answers that could not be read, any figure resting on a handful of questions, and the plain fact that the measure shows what assistants answered to our questions on the days we asked, not every buyer conversation.
What the board can decide: fund a named evidence gap, wait on a figure that rests on too few answers, or measure more questions or assistants. Put a cost beside each choice.
What stays off the page: a single blended score, a promised date, a promised number of recommendations, and any claim that pipeline came from AI assistants unless the data can show it.

Questions CMOs ask about AI visibility

Who is this page for?

It is for the person who owns the marketing budget and answers to the board for it: CMOs, heads of marketing and marketing directors at B2B technology companies. If you work in finance, content or demand generation, there is a page for your role as well.

Can we promise the board that assistants will recommend us?

No. Nobody outside the assistant makers controls what their assistants recommend, and answers vary from run to run. What you can promise is the work and the measurement: better evidence, measured on the same questions, reported with its limits.

How often should a CMO look at the figures?

Quarterly is enough for the board. The team doing the work may look monthly. Either way, compare the same questions on the same assistants, and report flat results as plainly as the ones that moved.

How long before we see a change?

There is no reliable published measure of how long a new page takes to appear in assistants' answers, and we do not offer one. Agree a measurement window before work starts, and read the result across repeated runs.

What should we ask an agency or a tool before buying?

Ask what exactly is measured, on which assistants and how many times, who does the work that follows, and how a change would be told apart from ordinary variation. Be wary of anyone who promises a date or a number of recommendations.

Do we need a tool before we start?

No. A small question set can be run by hand, and it is a good way to learn what you need. A tool earns its place once the set is large enough that repeated asking and counting become the bottleneck.

Stand out.
Get chosen.

Create your account in under a minute. Start with your first project and the questions that matter.

Get started Talk to our team

White paper aeroplanes with one yellow plane leading.