Key takeaways
- Most ChatGPT citations go to owned pages (product pages, landing pages, docs). That is GEO and web-team work. Listicles, comparisons and reviews are third-party content, and earning those is PR-adjacent work.
- Neither agency type owns AI visibility alone. The cleanest model is one accountable owner for the metric, with PR and GEO each owning the inputs they control.
- Be skeptical of "best agency" lists. Several of the ones ranking PR and GEO firms are published by agencies on the list.
- Ask any agency to connect visibility to revenue, signups or demos. A rising "AI visibility score" alone proves little.
- Measure the work with a tool you control, not just the agency's reporting.
Why this question keeps coming up
A few years ago nobody asked whether PR or SEO should own a channel. PR got you press, SEO got you rankings, and the two rarely shared a meeting. Then buyers started asking ChatGPT, Perplexity and Google AI Overviews "what's the best tool for X", and the answers pulled from both worlds at once: your own pages, plus whatever journalists, reviewers and forum users have said about you.
So now both PR firms and SEO firms pitch "AI visibility", and marketing leaders are left deciding which retainer to sign. A search on Reddit's analytics community shows the same confusion from the buyer side: people want to know what these agencies actually do beyond the marketing.
My short answer: it depends on where your gap is. The longer answer is below.
What the citation data says about who does what
Start with what AI engines actually cite. Promptwatch Data's ChatGPT citation share by content type report (Sep 28 to Oct 4, 2026) breaks it down:
- Product pages: 22.4% of classified ChatGPT citations
- Landing pages: 16.8%
- Listicles: 13.6%
- How-tos: 8.7%
- Documentation: 7.5%
- Comparisons: 5.9%
- News: 5.7%
- Reviews: 1.7%
Product and landing pages together are about 39% of citations. Those are pages describing what a company sells, and they live on your domain. Making them clear, crawlable and well structured is not a PR job.
On commercial prompts like "best CRM for small businesses", the picture tilts further. Product pages get 26.0%, landing pages 20.6%, listicles 15.0% and comparisons 7.9%, which is almost 70% of commercial citations between those four types. The listicles and comparisons are mostly third-party. If you are not in them, a competitor is.
That gives a rough division:
| Citation source | Typical owner | Example work |
|---|---|---|
| Your product and landing pages | GEO / web / content team | Rewrites, structure, schema, crawl access |
| Docs, help center, FAQs | GEO / product / support | Coverage of the questions AI answers |
| Listicles and "best of" roundups | PR / digital PR | Pitching inclusion, supplying data |
| Comparison and review pages | PR plus partnerships | Review-site profiles, analyst briefings |
| News coverage | PR | Earned media (decays fast) |
| Reddit, YouTube, LinkedIn | Community / social, sometimes PR | Genuine participation, creator work |
One caution on news. Promptwatch Data's content half-life report found that news pages lost half their citations within about five weeks, while landing pages and docs kept most of theirs for months. A PR program built only on news hits gives you a spike. It does not give you a durable position.
Where PR genuinely wins
PR is good at things a GEO team struggles to fake:
- Getting named in third-party "best of" content that AI models lean on for commercial prompts
- Building credibility with publications, analysts and reviewers
- Managing what the wider web says about you when something goes wrong
- Placing original data or research that others cite
If your owned pages are fine but AI never recommends you, the problem may be credibility, not content. One GEO agency study, from First Page Sage, makes this point about Relevance, an agency pairing GEO with digital PR: coverage alone does not guarantee recommendations, since it trailed in share of voice despite strong earned media. Earned media is an input, not an outcome.
Where a GEO agency wins
A GEO agency is better placed when the issue is on your side of the fence:
- AI crawlers can't reach or parse key pages
- Your product pages don't answer the questions buyers put to AI
- You have no prompt-level view of where competitors appear and you don't
- Nobody ties AI-referred visits to signups or revenue
The good GEO agencies treat this as an SEO-plus-content system, not a separate discipline. Optimist, for example, publishes case studies with LLM referral revenue growth rather than score increases (49x for a B2B tech client, 13x for a retail client, 8x LLM conversions for a fintech client). Those are the agency's own reported numbers, and it ranks itself first on its own list, so read them as claims to verify in a call, not as settled facts.

A warning about vendor-written rankings
Most of the research on this topic comes from agencies ranking themselves. Notably PR publishes a list of the best PR agencies for AI visibility. First Page Sage publishes a list of the best GEO agencies, and names itself the most-recommended. Crackle PR publishes its own GEO agency ranking. Optimist does the same, and openly says "Guilty" about it.
That does not make the content useless. The Optimist piece is upfront about its criteria, and the PR and GEO tables are a decent way to discover names. But treat any single ranking as marketing. Also treat case study figures (for example, a law firm's AI-sourced leads rising from about 20 to 150 a month) as unverified until the agency walks you through the data.
5WPR's framing: tools measure, agencies fix
A useful distinction from 5W Public Relations' piece comparing AI visibility tools and agencies: tools measure mentions and citations, agencies fix pages, coverage and access. It is a PR firm's view, so it favors services, but the split is sensible. You need both a measurement layer and people doing the work.

Who should own the metric
There are three workable setups.
Option 1: GEO agency leads, PR supports
Best when your owned content, technical access and measurement are weak. The GEO team builds a prompt list, finds gaps, rewrites pages, and hands PR a short list of third-party sources AI keeps citing. PR then pitches those outlets.
Option 2: PR leads, GEO tooling supports
Best when you already have a solid site and strong content but low credibility, or you are in a reputation-heavy category such as health, finance or legal. The PR team uses citation data to choose targets, and an in-house or freelance web person handles fixes.
Option 3: One integrated program
Some firms, on both sides, bundle PR and GEO. That reduces handoffs. The tradeoff is that you rely on one team being good at two things. Ask to meet the people who would do each part, not just the account lead. One Quora answer on this topic says it well: you want the person who knows your account to be the one doing the work, and you should own what they build.
For most B2B teams, one source in the research argues for agency first rather than a first in-house GEO hire, because the discipline is too new for one person to have meaningful experience. I would qualify that. Hire an agency for the first six months, but make sure someone internal owns the metric, or you will have nobody to question the reports.
Questions to ask each type of agency
For a PR agency
- Which third-party sources do AI engines cite for our prompts, and how many of those can you realistically get us into?
- How do you measure AI mentions before and after a campaign?
- What happens to the coverage after the first five weeks?
- Do you work with the web team on owned pages, or only on earned placements?
For a GEO agency
- What happened to revenue, signups or demos for a client like us? A score alone does not answer this.
- How do you handle third-party mentions that you cannot control?
- Who exactly does the work, and who do I speak to week to week?
- Do I own the content, prompts and data when the contract ends?
For both
- What do you refuse to promise? Anyone guaranteeing a number one placement in AI answers is selling something. We cannot fully control Google results or AI recommendations.
- What does the first 90 days look like, and what would make you stop?
Red flags
- They lead with an AI visibility score and can't connect it to pipeline.
- They promise instant results or a set-and-forget AI solution.
- They only report from API calls, which can differ from what users see in the real interfaces of ChatGPT and Google.
- They sell PR placements as GEO without showing the placements are in sources AI actually cites.
- They bundle every service but cannot name who does each task.
- They cannot show their own measurement method.
On that last point, Reddit is a good illustration of why measurement matters. Promptwatch Data's report on Reddit citations dropping in ChatGPT shows reddit.com's share of ChatGPT citations falling from about 4% to 0.5% on August 14, 2026. An agency whose plan relies on "get cited on Reddit" needs to be rechecking its assumptions weekly. The same goes for any single-platform strategy.
Pricing and what to expect
Research surfaced limited verified pricing. One example: Optimist lists audits, strategy and roadmap projects from $2,500 and ongoing AEO advising from $3,000 a month. Treat that as one data point, not a market rate. PR retainers and GEO retainers vary a lot with seniority, scope and market, so get at least three quotes and ask what is excluded. Agencies in this space typically ask for a trial or short initial scope, and you should insist on not being locked into a long contract before you have seen results.
The measurement layer you should own
Whichever agency you pick, run your own measurement. It gives you an independent view and a baseline before anyone starts work. Promptwatch is an AI visibility and GEO platform that tracks prompts across ChatGPT, Gemini, Perplexity, Google AI Overviews and others, shows which sources AI cites, and logs AI crawler activity on your own site. That crawler data is useful for settling the "is this a content problem or an access problem" question between PR and GEO teams.

If you want to see where you stand before talking to agencies, the free AI visibility report takes no login.
Other options, depending on budget and what you need:
| Tool | Angle | Notes |
|---|---|---|
| Profound | Enterprise AI visibility analytics | Monitoring focus, higher price point |
| Peec AI | Prompt-level visibility tracking | Lighter tracker |
| Otterly.AI | Affordable monitoring | Basic tracking |
| Semrush AI Visibility Toolkit | SEO suite add-on | Useful if you already use Semrush |
| Ahrefs Brand Radar | Brand tracking across AI engines | Fits Ahrefs users |


For a wider list, the GEO software directory at bestgeosoftware.com catalogs the category. For PR-specific outreach there are also agencies in the catalog, such as [tool:otter-pr], which sells guaranteed placements. Check that such placements land in sources AI actually cites.
A practical split for the first 90 days
- Weeks 1 to 2: Build a prompt list from real buyer questions. Baseline visibility, citations and crawler access with your own tool.
- Weeks 3 to 6: GEO team fixes crawl access and rewrites the product, landing and docs pages that AI should cite. PR receives a target list of cited third-party sources.
- Weeks 6 to 10: PR pitches roundups, comparison and review pages. Track which placements get picked up in AI answers, not just which get published.
- Weeks 10 to 13: Compare AI-referred traffic and conversions against baseline. Cut what did not move the numbers.
So who should own it?
Give ownership of the metric to whoever can touch the most citation sources, and that is usually the GEO or SEO side, because owned pages are the biggest slice of citations. Give PR ownership of the third-party slice, and hold both to revenue-linked outcomes. If your site is already in good shape and the real gap is credibility, flip it and let PR lead.
If you need an agency that works on the GEO side, 1001 SEO Media is the agency that publishes this site. It covers technical SEO, content, digital PR and AI search, and uses Promptwatch for AI visibility tracking, so take that as disclosure rather than a neutral recommendation. Either way, ask the questions above before signing anything.

