If you run an SEO or content agency, AI visibility is the rare new line you can sell to clients you already have, price above a normal SEO retainer, and deliver with the same team. This is the playbook for building it: what the service actually is, what goes in the monthly deliverable, how the leading shops price it, and the angle that makes existing clients say yes.
The reason it sells is that the money is already moving. Adobe, measuring more than a trillion visits across major retailers, reported AI-referred traffic to US retail sites up 393 percent year over year in early 2026, and that those AI-referred visitors convert better and spend longer on site than the average visit. On the other side, Pew found users click a traditional link on just 8 percent of pages that carry an AI summary, down from 15 percent without one. Clients are losing the click and gaining a higher-intent channel at the same time, and most of them have no one watching it.
What is new versus an SEO retainer
Most of your SEO craft carries over. Three parts of the job do not, and they are what a client is actually paying the premium for.
The measurement changes from rank tracking to answer auditing. Instead of a keyword’s position, you run a fixed set of buyer questions across ChatGPT, Claude, Gemini, Perplexity, and Google AI Mode on a schedule, and record whether the client’s brand was named, whether it was cited with a link, and who showed up instead. That produces a Brand Coverage rate and a Share of Voice against named competitors, per engine.
Per engine is not optional. The assistants disagree hard about who they will cite for the same question, so a single blended score hides the surface where your client is actually invisible. We measured that disagreement directly in the engine-overlap study, and it is the difference between a report that looks tidy and one a client can act on.
The work also moves off the client’s own site. In classic SEO you build authority through backlinks to pages you control. In AI answers, the assistant is usually assembling its response from third-party sources it already trusts: review sites, roundups, comparison pages, reference articles. Earning a mention on the three sources that already get cited for the client’s category moves more than one more post on the client’s blog. Backlink tracking becomes citation tracking.
The monthly deliverable
A retainer needs a repeatable shape a client can see value in every month. A defensible one looks like this:
- A baseline answer audit across the five assistants: Brand Coverage, Share of Voice against competitors, and the exact sources each answer cited.
- Crawl access and structured data: confirm the AI crawlers can reach the site, then ship schema and an llms.txt so the client’s own pages are legible when an engine does read them.
- Answer-first content built around the real buyer questions, not head-term keywords, so the pages actually match what gets asked.
- Off-site authority work: placements and corrections on the review sites, roundups, and reference pages the engines already cite for that category.
- One monthly report tying Brand Coverage and cited-source movement back to the client’s pipeline, not a vanity dashboard.
The crawl-access and structured-data steps are concrete and worth naming precisely rather than hand-waving as “be reachable”: allow the assistant crawlers in robots.txt, publish valid schema for the entity and its products, and keep an llms.txt current. Our AI visibility guide covers the access-first sequence if you want the full checklist to hand a junior on the team.
What the one-off audit actually contains
The audit is the product that opens the retainer, so build it well. The version that gets a client to act is not a dashboard export. It is a short, plain-language report built around the client’s own buyer questions, and it holds up because every claim is checkable. A format that works:
- A fixed set of the buyer questions a real customer would actually ask, run across all five assistants on one benchmark day and spot-checked over a couple of weeks so the pattern is not a one-day fluke.
- A clean split between being cited (the client’s page in the source list) and being named (the brand in the answer text). Clients are often a source and never a name, and that gap tends to be the whole story.
- Share of Voice against the named competitors who get recommended instead, so the client sees exactly who owns the shelf.
- A per-engine read, since the five assistants retrieve differently and the lever changes from one to the next.
- A split of the client’s search demand into the informational questions they may already win (how, where, when) and the commercial “best X” questions where buyers actually choose. Assistants tend to recommend operators from the commercial shelf, so that gap is usually what you are selling to close.
- One or two pieces of hard evidence: the specific answer where a competitor was recommended, and the page that earned it.
- A page-by-page plan naming the exact questions each page should answer, the off-site sources worth a mention, and a 90-day order of operations.
Two habits make it land. Name what the audit cannot do, plainly: no one can guarantee an assistant will recommend a brand, and saying so is what separates you from the tools overpromising in the client’s inbox. Then give the client a scoreboard: re-run the same questions in 90 days and compare. An audit a client can re-run is an audit that renews.
How the leading shops package and price it
Pricing for this work is still settling, so treat the numbers below as the range commonly quoted by agencies packaging it in 2026, not an audited market rate. Four models show up repeatedly, and they stack.
| Model | Commonly quoted range | Best for |
|---|---|---|
| One-time audit | 1,500 to 5,000 dollars | A low-risk first sale that proves the gap |
| Monthly retainer | 3,000 to 10,000+ dollars | The core recurring line |
| Add-on to an existing SEO retainer | +20 to 30 percent uplift | Expanding current accounts |
| White-label for other agencies | 40 to 60 percent markup | Reselling delivery capacity |
The four-model breakdown and the retainer ranges are laid out in more detail by agencies already selling it. One rule of thumb from that same crowd is worth repeating: a “GEO retainer” priced under about 1,500 to 3,000 dollars a month is usually relabeled classic SEO with a new cover page. If the measurement and the off-site citation work are not in the scope, the premium is not real.
The audit as a tripwire is the part most agencies underuse. A fixed-scope audit is a small, easy yes that ends in a report showing exactly where the client is missing from AI answers, which is the most direct path into the retainer conversation.
Sell it as retention, not a new logo
The fastest version of this business is not net-new clients. It is walking into an existing client’s next review with a report that shows their competitors getting named in AI answers where they are absent. That reframes AI visibility from an experiment into a hole in the work they already pay you for, and it is far easier to expand an account than to win one.
It also compounds across your book. The same weekly audit workflow, the same content and off-site motions, and the same report template run across every client brand once you have built them. One process, many accounts, which is the economics that make a specialist line worth standing up in the first place.
The number that closes the room
When a client asks whether they really have a problem, this is the honest answer from our own benchmark. Across about 26,400 AI answers we logged across the five assistants, the typical brand in a competitive category showed up in only about 13 percent of the answers about that category. The other roughly seven in eight named someone else or no one.
Read that as directional for a competitive category rather than a law of the whole web, since it is a sample of brands on questions where you would expect them to appear. Even so, most brands are absent from the large majority of AI answers about their own space, and being cited is the exception a client is currently leaving to chance. That gap is the service.
Where to take this next
If you are scoping the measurement side, our rundown of AI SEO tools compared covers what to look for so a client report shows cited sources and per-engine coverage, not just a mention count. And when a client wants to know why an assistant keeps naming a rival, why AI recommends your competitors is the piece to hand them before the strategy call. If you want to run the per-engine, per-client version of this workflow at scale, SurfacedBy offers trials on request for agencies evaluating it.
