• AI visibility

Guide · AI search

Who in my organization should own AI search visibility?

No single team can own it, because AI answers draw on your rankings, your own pages and what other sites say about you. The clearest published split gives SEO the job of getting found, content and product teams the pages themselves, and PR the third-party coverage. Someone also has to own the measurement, because AI visibility is easy to misread.

By Underneath Team Guide Updated

The short version

  1. A vendor-authored study recommends a three-way split: SEO drives whether you are found, content and product teams fix pages, and PR handles third-party outreach (Vishwakarma and colleagues (opens in a new tab)).
  2. SEO alone cannot cover it: 68.5% of the pages ChatGPT cited were not in Google’s top 100 for the question or for any of its searches (our AI citations and Google rankings study).
  3. Third-party coverage is a PR job with measurable weight: in 43.8% of its answers, ChatGPT ran a search aimed at a named publication, ranking or award (our hidden searches study).
  4. Your own facts need an owner: 18.9% of AI answers about local businesses stated a fact that differed from the business’s Google profile (our business facts study).
  5. Measurement needs a referee: a single ChatGPT answer showed only 57.8% of the brands that five answers to the same question named (our consistency study).

Is there one right owner for AI search visibility?

No: the work spans three existing functions, plus a measurement role. The most direct guidance comes from Vishwakarma and colleagues (opens in a new tab), who work at the software vendor Sprinklr. They write that “SEO teams drive retrieval rank, content/product teams own on-page fixes, and PR/partnerships own third-party outreach.”

Treat that as a recommended workflow, not a tested result. Their evidence is a simulated test in which AI assistants chose between two versions of a product review. Their internal pilot of the workflow reported only positive qualitative feedback, with no measured outcomes.

The split still matches what the wider evidence shows. AI answers depend on whether your page is found, what it says, and what independent sources say about you. Each of those already has a natural home in most marketing organizations.

What should the SEO team own?

SEO should own whether AI assistants can find and reach your pages at all. In the vendor study’s workflow, a brand absent from an answer’s citations has a finding problem, and the stated remedy is to improve SEO. In their tests, four factors mattered strongly in all six assistants: topic match, a stated price, a recent date, and being listed first.

But Google rankings explain only part of what AI assistants find. In our study of 80 US buyer questions, 68.5% of the pages ChatGPT cited were not in Google’s top 100 for the question or for any of the searches it ran. ChatGPT ran a mean of 3.7 searches per answer, none of them the user’s exact words (our hidden searches study). So the SEO brief widens: you need to rank for the searches assistants write, not only the ones people type. The SEO team will also want to know whether optimizing for ChatGPT hurts Google.

Technical access also needs a clear owner. In our crawler study, 51.7% of sites that block OpenAI’s training crawler still allow its search crawler. Most of those, 82.7%, never mention the search crawler at all, so the outcome is a default rather than a decision.

What should content and product teams own?

They should own what your pages say: facts, prices, specifications, dates and evidence. The vendor study’s owned-content advice is concrete: surface core topic terms early, add explicit price and key specs, include comparisons, keep dates current, and replace hedging with evidence. Of the 18 factors tested, 11 made a reliable difference in at least four of six assistants; formatting-only changes had little effect.

Consistency across your own pages is where this team earns its keep. In our pricing study, 61.9% of plan prices that AI assistants quoted were fully faithful to the official pricing page. Of the 64 prices that differed, 39 matched a figure on another page of the vendor’s own site.

The same pattern shows up for basic business facts. Where a business’s Google profile phone number did not appear on its own website, 30.6% of the numbers AI answers gave differed from the profile. Where it did appear, only 1.6% did (our business facts study). Conflicting facts on your own properties travel straight into AI answers.

What should PR and partnerships own?

PR should own what independent sources say about you, because AI assistants go looking for it. In 43.8% of its answers, ChatGPT ran a search aimed at a named publication, ranking or award. When a search named a source, the answer cited that source 44.0% of the time, against 8.1% when it did not (our hidden searches study). That is an association, not proof of cause.

Independent coverage was the strongest signal in our brand entity study. Each tenfold increase in the number of independent sites naming a brand in the cited pages went with 4.7 times the odds of being recommended. The vendor study’s third-party advice fits this: give reviewers missing data, enable independent testing, and support side-by-side comparisons.

Reviews belong here too, or with customer experience. In our reputation study, 88.0% of answers to “is this brand legit?” cited a review or complaint platform, and claims drawn only from review platforms were negative 56.5% of the time. In a simulated hotel test, a top guest rating raised the chance of being recommended by 31.6 percentage points (Baig and colleagues (opens in a new tab)).

Who should measure AI visibility and referee results?

One named person, ideally outside the teams being judged, because AI visibility is easy to overstate. Answers vary from run to run. In our consistency study, a single ChatGPT answer showed 57.8% of the brands its five answers to the same question named between them.

Traffic numbers need the same care. On one site studied by Watanabe and Nakayashiki (opens in a new tab), pages nobody optimized still grew 3.5 times in ChatGPT referrals, simply because ChatGPT grew. A team measuring its own work against last quarter would claim that growth as its result. The same site was used to test whether ChatGPT citations boost Google rankings.

Measurement does not have to be expensive. Researchers proposing independent audits of AI answers estimate that running them through public interfaces costs roughly $50–$300 at current pricing (Wen and colleagues (opens in a new tab)).

What should you do about it?

Give each part of the work a named owner and one shared scorecard. A practical setup:

  1. SEO: owns being found. That covers rankings for the searches AI assistants write, crawler access rules, and fixing pages that never appear in citations.
  2. Content and product marketing: owns on-page facts. One source of truth for prices, specifications, contact details and dates, kept identical across every page.
  3. PR and partnerships: owns independent coverage, reviews, rankings, awards and comparison pieces that AI assistants search for.
  4. Customer experience or operations: owns review volume and ratings on the platforms AI assistants cite.
  5. A measurement lead: owns a fixed list of buyer questions, asked repeatedly, with results compared against pages nobody changed.

Hold a short monthly review where the measurement lead reports and each owner takes one action. If you want outside help running that process, see our generative engine optimization service.

What does the research not tell us yet?

No study has tested which ownership model produces better results, so the split above is reasoned, not proven.

  • The three-way split comes from a vendor’s recommended workflow, backed by a simulated two-source test and an internal pilot with only qualitative feedback.
  • Our studies are observational snapshots of AI answers on given days. They show where citations come from, not what changes them.
  • The hotel finding uses fictional hotels in a controlled test, not live booking data.
  • Nobody has published evidence on budget splits, team sizes or reporting lines for this work.

Frequently asked questions

Should the SEO team own GEO?

The SEO team should own part of it, not all of it. In our study, 68.5% of the pages ChatGPT cited were not in Google’s top 100 for the question or its own searches, so rankings alone cannot explain visibility.

Is AI search visibility a PR job?

Partly, because AI assistants actively look for independent sources. ChatGPT ran a search aimed at a named publication, ranking or award in 43.8% of its answers in our study.

Who should fix wrong information about our company in AI answers?

Whoever owns your website and listings facts, usually content or operations. In our study, conflicting phone numbers across a business’s own sources went with far more mismatches in AI answers: 30.6% against 1.6%.

How do we know whether the work is paying off?

Assign one measurement lead and compare optimized pages with pages you left alone. On one site, untouched pages grew 3.5 times in ChatGPT referrals, so raw growth says little.

Sources

Free strategy call

Some questions are easier to answer about your own business.

Bring the one that matters most. On a free 30-minute call we’ll take a first look at it and send you a short written read afterward.

Ask about my businessHow we work