Agencies

How do agencies sell AI recommendations to clients?

Agencies sell the evidence, not a score: a before-and-after on the buyer questions AI already gets asked, the fixes applied to the client's site, and the AI visits those fixes produced. The report reads like work done, in the client's own words, not a dashboard number nobody can act on.

By Reflexa Technologies — the team building Reflexa, the agent that works on your site · November 10, 2026

In brief

This sits one level below brand intelligence itself, the practice of knowing what AI says about a brand before a client asks. An agency's job is turning that knowledge into something a client will keep paying for every month.

What does a client actually pay for?

A record of work, not a readiness score. The client meeting that starts this conversation is almost always the same question asked a different way: "what does ChatGPT say about us?" Answering it with a single number invites the next question, "so what do we do about it," which a score alone cannot answer. What a client can actually act on, and keep paying for, is a before-and-after: what the check found, what was fixed and when, the exact words that were published, and how much AI traffic each fixed page received.

Why it matters: a score changes for reasons nobody can explain in a client meeting. A list of fixes with dates and a traffic count next to each one defends itself.

How do you measure the before and after for a client?

Pick 10-20 real buyer questions in the client's category, ask each one across ChatGPT, Claude and Perplexity, and log three separate counts: Named, Recommended and Cited. Named means the brand appeared at all; Recommended means it was one of the top answers; Cited means the engine's sources listed the brand's own domain. The three move independently, so blending them into one score can hide a drop that is really a trust problem behind a rise that is really a content problem.

Worked example of the method: the Reflexa AI Visibility Index's September 2026 project-management edition found Asana recommended in 359 of 882 eligible answers, 41%, while it was cited on its own domain in a noticeably smaller share of those same answers, a gap between the two counts for the same brand (Reflexa AI Visibility Index, September 2026). The full method is published on How do I measure AI visibility?

Do this: run the same fixed question set every month for a given client, swapping questions between runs makes the before-and-after meaningless.

What goes in the "fixed" column of the report?

Whatever actually changed on the site, dated, with the words shown exactly as they were published. On a client's site that typically means crawler access rules, Organization and sameAs data, the sitemap and /llms.txt, plus page text, titles, descriptions and FAQ content that the client approved before it went live. Reflexa's own agent model runs one agent per client site, each with its own token, memory and history, so nothing from one client's account leaks into another's report.

Report columnWhat it should contain
FoundThe buyer question, the engine, and whether the client was named, recommended, or cited
FixedThe specific change, the date it published, and the exact words or setting involved
ResultAI visits to the changed page, before and after, counted per page

Bottom line: every row should name something specific enough that the client could point to it on their own site.

Can an agency white-label the report?

Usually not the dashboard itself. Reflexa states this plainly: the dashboard stays Reflexa's, and what an agency hands a client is the work and the numbers behind it, not a rebranded tool. None of the agency-pricing pages we checked across the field publish white-label as a feature either, so an agency's brand lives in the report it writes around the evidence, the meeting it runs, and the fixes it chooses to prioritize, not in the software's skin.

Do this: build the agency's own one-page summary template around the Found / Fixed / Result columns above, and treat the vendor's dashboard as the evidence locker behind it, not the deliverable itself.

How should this be priced against an SEO retainer?

The same way as any specialist addition: by the work it replaces, not a headline number. A developer or an SEO on retainer bills for the brief, the deploy and the check around a small change as much as the change itself, which is why a one-line title edit can still cost a half day of someone's time. No AI visibility vendor we have checked publishes a clean per-client-brand price either; Peec AI and Otterly.AI price by prompt volume, Profound and Scrunch AI bundle a flat entry tier, and Reflexa's Growth plan holds flat at €149/month with client sites not counted individually (reflexatechnologies.com pricing pages, October 2026). Ask a vendor to quote your exact client count before building a retainer line item around its entry price.

Proof: the full vendor-by-vendor pricing breakdown lives on Our agency pays per client brand. Which AI visibility tool should we switch to?

Rule of thumb: quote the AI visibility line item as a fixed monthly add-on per client, not a percentage of a retainer, since the work does not scale with the client's existing SEO budget.

What to do this week

None of this requires buying anything, and each step takes a few minutes per client:

Bottom line: the sellable version of this service is the report template, run consistently every month, not the tool behind it.

Keep reading

Frequently asked questions

How do agencies sell AI recommendations to clients?

They sell the evidence and the work: a before-and-after on real buyer questions (named, recommended, cited), the fixes applied to the site, and AI visits counted per page, reported in the client's own words each month.

What should an agency show a client instead of a score?

The actual AI answers to real buyer questions, what was fixed and when, and the AI traffic those fixes produced. A single score cannot be acted on, and it cannot tell a client whether a drop is a trust problem or a content problem.

Can agencies white-label an AI visibility report?

Not with most of the tools that measure it, Reflexa included. The dashboard stays the vendor's, so agencies brand the narrative and the before/after they hand over, not the underlying tool.

How should an agency price AI visibility work against an SEO retainer?

By the time it actually replaces or the work it does, not a headline price. Most AI visibility vendors do not publish a per-client rate either, so get the real bill at your exact client count before you quote one, the same discipline as any specialist line item added to a retainer.

Does every client need the same buyer questions tracked?

No. Each client's 10-20 buyer questions should be specific to their category and customers, not a generic set, because the entire report depends on asking the question a real buyer would actually type.

See what the report could already show a client.

The free check reads a site the way ChatGPT, Claude and Perplexity read it, with the raw answers and sources as evidence. 3 minutes, no card.

Run the free check →

Switched on by the agent: Organization schema in every page · sameAs profile links · på svenska