Every publisher serious about AI search has already built the apparatus: a set of prompts fired at ChatGPT, Gemini, Claude and Perplexity on a schedule, a record of which of its own pages get cited, and a view of which competitors keep appearing instead. Point that same apparatus at an advertiser's brand rather than your own titles and you have a diagnostic a chief marketing officer will pay for. That is the whole commercial idea behind a publisher GEO product, and it is why the category went from nothing to half a dozen live offerings inside about nine months. The reason it works is that the publisher is selling two things a GEO tool vendor cannot: the diagnosis, and the editorial authority that might actually move the answer.
What is a publisher GEO product?
A publisher GEO product is a paid service in which a media owner uses its own generative engine optimisation experience to improve or explain how a brand appears inside AI-generated answers. In practice the offerings that exist today bundle three things:
A visibility and sentiment audit. The publisher runs large prompt sets against multiple assistants and reports back on whether the brand appears at all, how it is described, which sources the models are leaning on, and where the description diverges from the brand's own positioning.
AI-oriented branded content. The publisher writes and publishes content designed to be retrieved and cited rather than read: comparisons, listicles, reviews, structured explainers, sometimes video. It sits on the publisher's high-authority domain, and often on third-party surfaces such as YouTube as well.
Ongoing measurement. Tracking citation volume, share of voice and sentiment over subsequent weeks and months, with the explicit understanding that models change underneath you.
The category label is unsettled. Digiday and most vendors say GEO. Edward Cowell, global vp of SEO and GSO at WPP Media, argues for "generative search optimisation" or GSO instead, on the grounds that treating AI visibility as separate from search is precisely the mistake that damages sites. That naming argument matters commercially, because it is the fastest way to tell a serious offering from an opportunistic one.
Which publishers are already selling this?
Future is furthest along. It built a proprietary tool, Future Optic, launched in beta in November 2025, initially because third-party AI visibility tools were built for brands rather than publishers. As Simon Collis, Future's svp of content strategy and audience development, told Digiday, publishers need to know "where the information is coming from in the first place, which specific pages are feeding the models, and how our content is actually influencing the answer". Future now sells Optic as part of branded content packages to clients in telecoms, CPG, luxury and beauty. Its Samsung campaign drove 28% growth in citations from Future sources over three months, with an uplift in mentions of between 23% and 33% and 4,754 LLM citations recorded at the end of August, per the company's FY2025 results presentation. Future declined to say how many brands are using it or what it charges.
Time launched a variant built around sentiment rather than citation count, announced by chief operating officer Mark Howard at the Digiday Publishing Summit in Vail in March 2026. Working with Mobian, Time analyses a brand's own advertising for tone and positioning, generates up to 100 prompts across ChatGPT, Claude, Gemini and Perplexity, and scores the delta between what the brand says about itself and what the assistants say about it. It then sells Time-branded content to close that gap. Mobian's analysis of 750 brands found 17% of citations in AI search were either misaligned with the brand's positioning or factually inaccurate, according to co-founder and CEO Jonah Goodhart, who described it as a brand safety and suitability application. Howard declined to give pricing, and said it falls inside Time's existing branded content packages.
Germany's BCN, the joint-venture commercial arm of Hubert Burda Media, Funke and Klambt, launched GEO Brand Impact in June 2026: an audit built on third-party tools including Peec AI and Rankscale, followed by AI-optimised branded content placed across roughly 300 specialist titles including Chip, Elle and Grazia, with the JV claiming reach of around 63 million users. Stefan Betzold, BCN's chief digital product officer, was explicit that his Brand Stories team is now being trained on "how to write for a bot", and equally explicit that this is a brand awareness product priced at the premium end of the branded content range, not a performance buy.
Forbes, Axios and The Washington Post occupy the earlier stages. Forbes, which Profound has ranked as the most-cited publisher in ChatGPT, is fielding client questions about how their branded and native content performs against campaign-aligned prompts; chief innovation officer Nina Gould said in February 2026 that Forbes was not packaging it as a standalone product but saw clear opportunity. Axios, cited among the top sources across four separate studies including Muck Rack and 5W, was in early conversations with advertisers as of July 2026, per CRO Jacquelyn Cameron. The Washington Post hired Kyle Sutton as its first head of AI discovery, and CRO Karl Wells has said AI referral traffic grew 16% over six months and that visitors arriving from LLMs convert to subscribers at four to five times the rate of traditional search.
Why are brands buying it?
Because the anxiety is real and it is documented. In WordPress VIP's Future of the Web 2026 research, 74% of 800 enterprise decision-makers and CMOs said AI discoverability and attribution were a main or significant priority, with teams already spending an average of more than two working days a week trying to improve them. The same report found that, on average, 60% of enterprise audience reach now comes from third-party platforms brands do not own or control.
Betzold's framing of the buying trigger is the clearest one on the record: "Just imagine you're a CMO. You try to find out what's happening with your new product in AI search... and then you find out you're not visible. There is not yet a proper market, so who do you ask?"
The demand is now inbound rather than pitched. Time received multiple RFPs in a single week in July 2026 seeking solutions to improve how a brand shows up in AI platforms. Publishing executives told Digiday the products exist because brands came to them first.
What is a publisher actually selling here?
This is the part most coverage skips, and it is the part that determines whether the product survives its second year. A publisher GEO product contains three commercially distinct things, and they carry very different risk.
The diagnostic is real and defensible. Nobody disputes that you can measure, at a point in time, whether a brand appears in a set of AI answers and how it is characterised. A publisher can run this credibly. The methodological caveats are large but honest.
The authority signal is real but unquantified. The genuine publisher advantage is that assistants demonstrably lean on trusted, well-structured sources, which is why the licensing market exists at all. Wrapping a brand in credible coverage on a high-authority domain plausibly improves its chances of being retrieved. Rita Steinberg, vp of media at FUSE Create, put the limit precisely: publishers can improve the conditions for visibility, but cannot credibly guarantee that a specific piece of content will appear in an AI-generated answer. She would test it, she said, but would "push hard on measurement, methodology and proof before paying a major premium".
The outcome is not yet sellable at all. Goodhart was candid that it remains unclear whether AI systems weight branded content differently from editorial, and that whether any of this changes how a model perceives a brand is "way too early to know".
A publisher that sells the first two and is honest about the third has a durable product. A publisher that sells the third has a refund conversation coming.
Why this cannot be priced as a performance buy
There are no reliable click-throughs from AI answers, so last-click attribution has nothing to attach to. That single fact forces the commercial model: consultancy and analytics fees plus premium branded content rates, with results assessed over weeks and months.
Worse, the underlying metric is not standardised. Every analytics vendor uses a different dataset and methodology, and their numbers disagree. Howard's comparison is the useful one: it resembles ad viewability a decade ago, when every verification vendor had its own measurement system and classification and nothing was interoperable until the industry standardised it. Until an equivalent standard arrives, no publisher can anchor a rate card on a number a buyer can independently verify.
Time's workaround is instructive and worth copying. Rather than leading with contested citation rankings, it uses its own AI bot traffic as the proxy metric in client conversations, on the basis that heavy crawler demand for your content is direct evidence that models want it. Time reports sitting in the 98th percentile for AI bot activity among the nearly 7,000 publisher sites in TollBit's network. Server logs are first-party, countable and hard to argue with, which is exactly what a contested market needs.
There is also a cautionary precedent on the agency side. Agencies including Digitas, Jellyfish and Botify launched AI visibility tools ahead of publishers, and several found the tools hard to turn into sustainable revenue because clients saw limited results from insights alone. Insight without an execution layer does not renew. This is the structural reason publisher offerings bundle content: the content is the part the client cannot buy anywhere else.
The credibility risk publishers are walking into
The GEO vendor market has a reputation problem, and publishers entering it inherit that problem by association. Kevin Indig's Growth Memo counted $227 million of investment into AI visibility tracking services. Cowell said he receives GEO vendor pitches every 24 to 28 hours. Publishers told Digiday in July 2026 that a noticeable share of GEO vendors are closer to snake oil than partnership, pitching top-citation guarantees with no credible method behind them.
Three tactics are actively dangerous and no publisher should let them near a client deliverable.
Mass AI-generated content. Lily Ray, vp of SEO strategy and research at Amsive, has documented vendor case studies showing AI visibility gains alongside organic traffic collapse, including one company that lost 66% of its traffic. Microsoft's Bing team published guidance in December 2025 on duplicate content and AI search visibility for the same reason.
Prompt injection. Embedding instructions on a page that only the model sees, typically via a summarise-with-AI control, telling it to rank the company first. Microsoft named this in February 2026 as AI recommendation poisoning and treats it as an attack.
Selling GEO as a replacement for SEO. Jeremy Moser, co-founder and CEO of uSERP, put the test plainly: "The 80/20 of GEO is simply just good traditional SEO. If a GEO service does not openly tell you that success in AI visibility is 80 percent good fundamental SEO, they are selling you snake oil."
For a publisher the exposure is asymmetric. A tool vendor that oversells loses a client. A publisher that oversells puts its editorial authority, which is the only asset making the product work, on the line.
What a publisher needs before selling this
Run it on yourself first, for at least two quarters. Every credible offering in the market started as internal defensive work. Future built Optic for its own titles. Time and Forbes both sold from a position of documented citation strength. Without your own before-and-after you have a deck, not a product.
Instrument your own logs, not just a third-party dashboard. Your crawler and retrieval data is the one number in this category that is first-party and countable. It is also the number you already own and mostly are not reporting.
Write the disclosure into the contract. State explicitly what the product influences and what it cannot control. This is the clause that protects the masthead when a campaign does not move the answer, and buyers such as Steinberg are already asking for it unprompted.
Reconcile it with your crawler policy. There is an obvious tension in charging a brand for AI visibility while blocking the crawlers that would deliver it, and 56.4% of news publishers now block at least one AI crawler in robots.txt, per HasData. Selective access, in which routine coverage is open and exclusives are not, is the position most publishers are converging on. Decide it deliberately rather than letting two departments answer differently.
Price it as a retainer, not a one-off. Howard's point that this shifts branded content from single executions to consecutive releases over a longer period is the commercially important one. Models change monthly. A one-off audit dates in weeks; a standing relationship is what the volatility actually justifies.
Where selling visibility ends and monetising retrieval begins
A publisher GEO product monetises expertise about the AI layer. It does not monetise the AI layer itself. Those are separate revenue lines and it is worth keeping them separate on the P&L, because the second one behaves like inventory rather than like consultancy.
That second line is where blankspace operates. blankspace detects Live Search Agent traffic at the CDN edge, runs an auction and injects contextual brand facts into the response the agent is about to synthesise, with same-day attribution. It is a different transaction from selling a brand advice on how to be cited: it is charging for the retrieval event that is already happening, on inventory that never renders JavaScript and therefore never appears in a conventional ad call. The broader point holds regardless of vendor. A publisher building a GEO practice is building a commercial team fluent in exactly the vocabulary needed to sell the retrieval layer too, and most of the diagnostic work needed for one supports the other.
Frequently asked questions
Is a publisher GEO product the same as an AI visibility tool?
No. An AI visibility tool such as Profound, Peec AI or Rankscale samples prompts and reports what the assistants said. A publisher GEO product usually incorporates one of those tools for the audit, then adds the thing a tool cannot supply: content published on a high-authority domain that might change the answer. Publishers are selling the execution layer, and the agency experience suggests insight alone does not sustain revenue.
Can a publisher guarantee my brand appears in ChatGPT or Gemini?
No, and any offering that promises it should be treated as a warning sign. Retrieval depends on authority, relevance, source quality, recency, existing search signals, citations, the wording of the user's prompt and the model itself, most of which no publisher controls. The credible claim is that the publisher improves the conditions for visibility.
How much does it cost?
No publisher has put a rate card on the record. Both Time and BCN have said it sits inside their existing branded content pricing, with BCN placing it at the premium end of that range and framing it as a longer-term retainer rather than a one-off execution. Expect it to be quoted as branded content plus a consultancy or analytics fee.
Does selling GEO replace lost search referral revenue?
Not at anything like the required scale, and publishers making the argument have not claimed otherwise. Ozone's benchmarking showed publisher ad supply falling by up to 40% in Q2 2026 as zero-click search reduced traffic to open-web sites. A branded content line item does not close a gap of that size. The honest case for building one is diversification and commercial relevance in the channel where discovery is moving, not replacement.
Should a smaller publisher attempt this?
Only with a defensible niche. The publishers succeeding here have either enormous general authority, such as Forbes and Time, or deep vertical authority, which is the whole basis of BCN's roughly 300 specialist titles and Future's decision to track 13 subverticals rather than broad categories. A specialist publisher that genuinely dominates retrieval for a narrow topic has a better product than a mid-sized generalist, because the prompt set it can win is one it can name.