Are Your PR Services Showing Up in ChatGPT? A Visibility Audit for Brands

Ask ChatGPT for the best PR firm for a Series B fintech, and you'll get a short list in about four seconds. Five or six names, a sentence on each, maybe a link or two. The buyer reads it, opens two of those websites, and the shortlist is done before anyone has filled out a contact form.

If your firm isn't on that list, you never hear about the deal. There's no lost-lead notification for an AI answer that skipped you.

That's the problem an AI visibility audit solves. An AI visibility audit for PR services is a structured test of how often, how accurately, and how favorably ChatGPT and other AI assistants mention your firm when buyers ask the questions that lead to hiring a PR partner. It measures presence, not rankings, and it traces which sources the model leans on so you know what to fix.

Below is the audit we'd run on any agency or firm offering PR services, and it works just as well for in-house comms teams. You can do most of it in an afternoon with a spreadsheet.

Why ChatGPT visibility matters for PR services right now

Buyers have already moved. Responsive surveyed 350 B2B buyers in October 2025 and found that generative AI has overtaken traditional search for about a quarter of them when researching vendors. Among U.S. buyers, 48% use GenAI for vendor discovery. At companies with more than 2,000 employees, it's 42%. Those are exactly the accounts most PR firms want.

And PR has an odd advantage here. AI models trust what others say about you far more than what you say about yourself. Muck Rack's May 2026 What Is AI Reading? report, which analyzed more than 25 million links cited by AI tools, found that 84% of citations came from earned media. Journalism alone made up 27% of cited links. The same report found ChatGPT cites sources in 96% of its responses, averaging about five citations each.

So the firms whose entire business is earning coverage should be dominating AI answers. Many aren't. The cobbler's children problem is real: agencies spend their days landing placements for clients and rarely do it for themselves. Their own websites carry the weight, and AI models discount owned content.

How does ChatGPT decide which PR services to recommend?

There's no secret ranking algorithm you can reverse-engineer, but the inputs are visible enough to work with.

When ChatGPT answers with web search on, it pulls live pages and cites them. When it answers from its trained knowledge, it draws on patterns from its training data about your firm, your competitors, and the category. In both cases, the same thing wins: repeated, consistent, third-party descriptions of who you are and what you're good at.

A PR firm that shows up in trade publication roundups, award shortlists, journalist quotes, Wikipedia-adjacent sources, directories like Clutch, and industry analyst coverage gives the model plenty of corroboration. A firm whose only description of itself lives on its own services page gives the model one biased source, and the model treats it that way.

Specialization matters too. "Full-service PR agency" means nothing to a language model. "Crisis communications firm for healthcare systems" is something it can match to a prompt.

The PR services visibility audit, step by step

Step 1: Build a prompt set that mirrors how buyers actually ask

Skip the vanity prompt ("What is [Your Firm]?"). Buyers don't ask that until they already know you. Write 20 to 30 prompts across four intent types:

1. Category discovery: "Best PR agencies for B2B SaaS companies," "top crisis PR firms in the US," "PR services for a hospital system facing a lawsuit."
2. Problem-led: "How do I fix negative press coverage about my CEO," "who can help us respond to a data breach publicly."
3. Comparison: "[Your Firm] vs. [Competitor]," "is [Competitor] good for startups."
4. Brand direct: "What is [Your Firm] known for," "is [Your Firm] reputable," "how much does [Your Firm] charge."

Mix in geography and industry modifiers that match your real pipeline. If 60% of your revenue is healthcare, half your prompts should be healthcare.

Step 2: Run each prompt more than once

This is where most DIY audits go wrong. They run a prompt once, screenshot it, and call it a finding.

AI answers aren't stable. In a study of nearly 3,000 runs across ChatGPT, Claude and Google's AI, SparkToro's Rand Fishkin and Gumshoe.ai's Patrick O'Donnell found the chance of getting the same recommendation list twice was under 1 in 100. Getting the same list in the same order was closer to 1 in 1,000. Brands that were genuinely strong still appeared in 60% to 90% of responses for a given intent, even as their position bounced around.

The lesson: track how often you appear, not where. Run every prompt at least five to ten times, in fresh chats, logged out or in a clean session, with web search both on and off. Record each result.

Step 3: Score four things, not one

Presence is the headline number, but it hides problems. Log these for every run:

1. Mention rate: Did your firm appear at all? Your score per prompt is mentions divided by runs.
2. Accuracy: Did ChatGPT describe your services correctly? Watch for outdated offerings, wrong locations, merged entities, and old leadership names. A model confidently saying you do media buying when you dropped it in 2023 is worse than no mention.
3. Sentiment and framing: Were you recommended, listed neutrally or caveated ("some clients report...")? One old negative review thread can follow you into every answer.
4. Citation source: When ChatGPT linked something, what was it? Your site, a trade publication, a directory, a Reddit thread, a competitor's blog?

A simple spreadsheet works: prompt in column A, run number in B, then one column for each of the four scores.

Step 4: Trace the citations back to their source

Citations are the most actionable part of the audit. They show you the shelf the model is pulling from.

Sort every cited URL by domain. You'll usually find a handful of publications, directories and listicles doing most of the work. If a competitor's name keeps appearing alongside a link to a "Top 15 PR Firms" roundup on an industry blog, that roundup is doing more for their AI visibility than their entire website.

Then ask three questions about each frequently cited source. Are you on it? Is your description there accurate? Could you get on it through outreach, an award submission, a contributed piece, or a corrected listing?

Recency matters here. Muck Rack's data shows more than half of journalism citations come from articles published in the past 12 months, with citation volume dropping sharply after the first six. A great 2022 profile in a trade outlet probably isn't carrying you anymore.

Step 5: Run the same prompts for three competitors

Your mention rate means little in isolation. A 30% rate is terrible in a category where the leader sits at 85% and excellent in a fragmented market where nobody clears 25%.

Pick the three competitors you lose deals to most. Run your full prompt set for their brand-direct questions, and log their appearances in your category prompts. The gap between you and them, prompt by prompt, is your priority list.

This is also where you find the ugly surprises. A competitor you barely consider may be dominating AI answers because of one well-placed Wikipedia mention or a steady run of bylines in a niche publication.

Step 6: Check whether ChatGPT can even read your site

Earned media carries most of the weight, but your own site still needs to be legible. Two quick checks:

1. Crawler access: Open your robots.txt and confirm you aren't blocking OAI-SearchBot, the crawler OpenAI uses to surface sites in ChatGPT search. Overzealous security plugins and copy-pasted "block all AI bots" rules catch a surprising number of agencies.
2. JavaScript rendering: If your services pages load their copy through JavaScript, AI crawlers may see an empty shell. View the raw page source. If your service descriptions aren't in the HTML, fix that before anything else on this list.

Ongoing AI visibility monitoring can automate much of steps 2 through 5 at scale. NetReputation, for example, tracks how often it appears for prompts like "who is the best reputation management company?" across ChatGPT, Gemini and other assistants, and uses shifts in that share to decide where to push earned coverage next. But a manual audit is the right starting point. You'll understand the numbers better once you've seen the raw answers.

What the audit usually reveals for PR firms

After running this on enough communications brands, the same patterns keep showing up:

1. Strong brand-direct answers, weak category presence: ChatGPT knows who you are when asked by name, but never volunteers you in "best PR firm for X" answers. This means you have recognition without positioning. The model doesn't connect you to a specific problem.
2. Accurate presence in only one niche: You appear for "tech PR" but vanish for "crisis communications," even though crisis work is half your revenue. Your earned coverage over-indexes on one practice area.
3. Outdated service descriptions: The model lists offerings you no longer sell or misses ones you launched recently. This usually traces back to stale directory profiles and old press.
4. Competitors winning on one source: A single listicle or roundup accounts for a big share of a rival's mentions. Get on that list, and the gap narrows quickly.
5. Negative framing from a forum thread: One unresolved complaint on Reddit or a review site gets cited repeatedly, coloring every answer that mentions you.

How to fix weak ChatGPT visibility for your PR services

Fixes fall into two buckets: what others say about you, and what you say about yourself. The first bucket matters more.

Earn coverage that names your specialty

Pitch yourself the way you pitch clients. Contributed articles, expert commentary, podcast appearances, and award submissions all count, but only if they describe you specifically. "Jane Smith, founder of a PR firm" teaches the model nothing. "Jane Smith, whose firm handles crisis communications for regional banks" does.

Prioritize the publications that showed up in your citation trace. If ChatGPT keeps citing three trade outlets for your category, those three are your pitch list for the next quarter.

Clean up every third-party profile

Update Clutch, G2, LinkedIn, Crunchbase, industry association directories, and any "top agencies" lists you're already on. Make the service descriptions match each other and match your website. Consistency helps a model gain confidence that a fact is true.

Rewrite your services pages for extraction

AI models pull clean, quotable statements. Open each services page with a one- or two-sentence answer to "what does this firm do, for whom?" Add real specifics: industries served, team size, typical engagement length, named outcomes you're allowed to share. Cut the paragraphs about being "passionate partners in storytelling."

Add an FAQ section that answers the questions buyers actually type into ChatGPT, in plain language, with each answer able to stand alone.

Address the negative sources directly

If a complaint thread keeps getting cited, respond publicly and professionally where the platform allows it, and work to resolve the underlying issue. Then build enough fresh positive coverage that the old thread is one source among many rather than the loudest one.

How often should you rerun a ChatGPT visibility audit?

Quarterly is the right rhythm for most PR firms. AI models update their behavior, search indexes refresh, and your earned coverage changes month to month. A quarterly rerun of the same prompt set lets you see whether your outreach is moving the mention rate.

Run an extra audit after any major event: a rebrand, a new practice area launch, a leadership change, or a piece of negative press. Those are the moments when AI answers drift fastest, and when stale descriptions do the most damage.

Keep the prompt set mostly fixed between audits so the numbers stay comparable. Add new prompts as your services change, but don't swap out the core.

Frequently asked questions

How do I check if ChatGPT recommends my PR services?

Write 20 to 30 prompts a real buyer would use, such as "best crisis PR firms for healthcare," and run each one five to ten times in fresh chats. Record how often your firm appears, whether the description is accurate, and which sources ChatGPT cites. Your mention rate across runs is the number to track.

Why doesn't ChatGPT mention my PR firm even though we rank on Google?

Google rankings and AI mentions draw on different signals. ChatGPT leans heavily on third-party sources like trade publications, directories, award lists, and industry roundups. If your visibility comes mostly from your own website, the model may not find enough independent corroboration to feel confident recommending your firm by name.

What is the difference between SEO and AI visibility for PR services?

SEO aims to rank your pages in search results. AI visibility aims to get your firm named and accurately described inside AI-generated answers. They overlap, since crawlable pages help both, but AI visibility depends more on what other sites say about you and less on your own keyword targeting.

How long does it take to improve ChatGPT visibility?

Expect three to six months before a meaningful shift in mention rate. Fixing crawler access and directory profiles can help within weeks, but earned coverage takes time to publish and get picked up by AI models. A steady rhythm of monthly placements will do more for you than one big feature that goes stale.

Can I pay to have my PR services featured in ChatGPT?

Not inside the answer itself. OpenAI has started showing labeled ads in ChatGPT, but it says ads are kept separate from responses and don't influence what the model recommends. Organic mentions come from the sources ChatGPT reads and cites, so earned media, accurate listings, and clear service pages remain the levers.