For marketing agencies
AI visibility for marketing agencies: six workflows for running an AI search practice
Clients are asking what ChatGPT, Gemini, and Google AI Overviews say about them. These six workflows turn that question into a practice you can sell, deliver, and renew, from the first visibility report through monthly reporting and competitive displacement.
Last updated
6 of 28 guides in the Spyglasses workflow series.
Your clients are asking about AI. The answer can't be a screenshot.
Clients started asking what ChatGPT says about them, and a lot of agencies answered with a screenshot. That works once. It doesn't work as a retainer, because the same prompt gives a different answer tomorrow and the client notices.
The way out is to stop treating the final answer as the thing you optimize. Answers vary by session. The steps before the answer don't. AI searches the web, ranks what it finds, reads pages, and scores them for relevance, and those steps follow published rules you can work against. Track the answers to prove the work; do the work on the pipeline.
Each workflow below is a full article with the steps, the order, and what to hand the client at the end. Read them in sequence if you're building the practice from scratch, or jump to the one you need this week.
Key takeaways
- AI answers vary by session, so a single screenshot isn't a measurement.
- The retrieval steps behind the answer are repeatable, and that's where the work happens.
- An AI practice is a sequence: prospect, baseline, track, report, then displace.
- Client-ready exports matter as much as the data; the deliverable is what gets renewed.
Win the work
The pitch is easier when the prospect's own numbers are on the screen.
Set the account up
Two workflows, run once per client, that everything else depends on.
Run the retainer
The recurring work, plus the two engagements clients ask for by name.
What you'll use along the way
The workflows stand on their own, and every one of them names the tool it leans on.
Start with a prospect's numbers
Run a free preview on any domain and see what the major assistants say about it today. It takes a few minutes and it's the same report you'd hand a client.
Other roles, same series
The workflows overlap more than the job titles do. These hubs cover the rest of the set.
AI visibility guides for PR and communications teams
PR and communications →AI visibility guides for in-house marketing teams
In-house marketing →AI visibility guides for content teams
Content teams →Frequently asked questions
How is this different from a prompt tracker?
Prompt trackers report where you stand, which is useful and is table stakes at this point. The gap is knowing what to do next. Every workflow here ends in a change you can make and a way to check whether it worked.
What do I show a client in month one?
A baseline report with share of voice, brand consistency, the competitor set, and a per-prompt breakdown, plus the prompt set you'll track from then on. Start the tracking in the same month so month two has something to compare against.
Where do I start with a client who has never measured AI visibility?
Run the baseline workflow first. It gives you one defensible number, a corrected brand snapshot, and a competitor set, and everything you report later gets measured against it.
How do I show a client that the work moved the number?
Log each piece of work as an annotation on the trend line, then let the monthly report put the work next to the movement. The story is the sequence, not the metric on its own.