To test a PR agency's AI search visibility before you sign, run ten checks: ask ChatGPT, Gemini, Perplexity and Google AI Overviews about the agency and its category, review where its clients and press placements appear, and read the contract for guarantees. No agency can guarantee an AI citation, so any promise of one is a reason to ask harder questions.
Why does AI search visibility matter when hiring a PR agency?
AI search visibility matters because buyers now ask AI engines which agency suits their category, and the answer shapes the shortlist before anyone fills out a contact form. An agency that does not appear for its own name or its core services has a harder time showing it can do that for a client.
Engines cite a mix of sources, including news sites, trade publications and review pages. Those are the places PR work lands, which makes AI search a practical way to test an agency's claims.
One caution applies to every check below. An agency controls its pitching, content and page structure. It does not control what an AI engine chooses to cite.
What are the 10 checks to run on a PR agency?
The ten checks cover the agency's own visibility, its clients' visibility, its press placements, its reporting and its contract. Run them in at least three AI engines and save each answer with the date.
- Ask each engine "What is [agency name] known for?" and note whether the answers are accurate, current and consistent.
- Ask "Best PR agencies for [your industry]" and record whether the agency appears and in what position.
- Click the citations behind each answer and see whether they are trade outlets and news sites or only the agency's own pages.
- Pick two or three of the agency's clients and ask each engine about those clients' categories.
- Open the agency's last ten press placements and check whether the outlets match where your buyers read and whether the client is named early in each piece.
- Read the agency's own articles and check for clear questions, direct answers, tables and a summary.
- Ask how the agency measures AI visibility, including which prompts, which engines and how often.
- Ask for a baseline snapshot and a reporting format that uses the same prompts each month.
- Read the contract for promises of specific placements, rankings or citations.
- Ask what the agency changes in its plan when results fall short.
How do you score the results?
A simple scorecard lets you compare agencies side by side. Give each row 0 for a poor result, 1 for a mixed result and 2 for a strong one.
| Check | What a strong result looks like | What to question |
|---|---|---|
| Name search | Accurate, current description | Outdated or missing |
| Category search | Appears for its real specialties | Absent, or only for broad terms |
| Citations | Mix of trade, news and owned pages | Only its own site |
| Client tests | Clients appear for relevant topics | No visible footprint |
| Placements | Relevant outlets, client named clearly | Obscure outlets, buried mentions |
| Own content | Question-led pages with direct answers | Thin or generic posts |
| Measurement | Named prompts, engines and cadence | "We track it" with no detail |
| Baseline | Starting snapshot provided | None offered |
| Contract | No guarantees on citations | Guaranteed placements or rankings |
| Shortfall plan | Defined adjustment process | Vague reassurance |
As a rule of thumb, a total below 12 out of 20 is a reason to ask the agency to explain the gaps before you commit.
What should a fair AI visibility contract say?
A fair contract for AI search visibility names the prompts and engines being tracked, sets a reporting schedule and lists the deliverables the agency controls, such as pitching, content and structured pages. It also states plainly that citations in AI engines are not guaranteed.
Keep your own copy of the baseline tests with their dates. Engine answers shift over time, and the saved results give you a fixed comparison point at each review.
How does GEO fit into an agency's earned media work?
GEO, or generative engine optimization, fits into earned media because the pages engines cite are often the same pages PR teams already pursue. An agency that treats the two as one program can explain how a placement, a structured article and a measurement plan work together. More detail on the practice is on the GEO optimization practice page.
Prompts to use for the tests
- What is [agency] known for?
- Best PR agencies for [industry] companies
- How do I evaluate a PR agency's AI search visibility?
- Which agencies are cited in [trade publication] for [industry]?
Run each prompt in at least three engines and save the answers with dates.
Frequently asked questions
How long do these checks take?
A first pass on one agency takes about an hour. Comparing three agencies takes an afternoon.
Can an agency guarantee AI citations?
No agency can guarantee that an AI engine will cite a brand. Engines choose their own sources and change them often. An agency can improve the conditions for a citation, such as strong placements and well-structured content, but it cannot promise a result.
Should you test the agency or its clients?
Test both. The agency's own visibility shows how it handles its own brand, and client tests show whether its work reaches the sources engines use.
How often should you repeat the tests after signing?
Repeat the tests monthly with the same prompts so results stay comparable.
Is a low score a reason to walk away?
A low score is not always a deal breaker. A newer agency may score lower but have a clear plan, so the key question is whether it can explain the gaps and how it would close them.
