Inside the GEO cottage industry

The race to show up in AI-generated summaries has sparked a new vendor boom — and not everyone is convinced it’s legit.

A cottage industry of vendors has mushroomed over the past year, each one promising to optimize publisher content for AI search visibility – what’s known as generative engine optimization (GEO).

Inboxes are being flooded with outreach from these companies. Edward Cowell, global vp of SEO and GSO at WPP Media, said he receives pitches from GEO vendors every 24 to 28 hours. Kevin Indig’s SEO-focused newsletter, Growth Memo, found that investors have poured $227 million into AI visibility-tracking services.

“VC money is flowing into anything and everything AI and GEO. I believe the reason why so many of these firms and agencies are either cropping up for the first time or repositioning to AI search is simply to capture the gold rush,” said Jeremy Moser, co-founder and CEO of SEO agency uSERP.

He pointed to an “SEO/GEO agency” that raised VC from well-known Silicon Valley funds, as an example. The company has large corporate logos on their site, but the founders have no visible expertise in SEO on their LinkedIn, and their main offering is $299/mo, publishing 30+ articles on autopilot. “That’s it. The entire strategy is publishing fully AI-written articles at scale,” said Moser. “This is a surefire way to tank your domain organically as it goes directly against Google’s spam policies.”

It’s a familiar pattern: publishers have already lived through the SEO and ad tech vendor explosions. Now, the GEO vendors have arrived. 

Whether they are selling solutions or snake oil is still an open question.

The issue isn’t whether GEO matters. It most certainly does. The issue is that publishers and brands are being asked to spend money optimizing for a black box whose outputs can’t be measured in any commercially meaningful way yet. 

SEO veterans are seeing straight through much of the hype. Several told Digiday that many of these GEO tactics closely resemble traditional SEO practices repackaged for AI. Or, even worse, these services risk hurting a site’s search visibility or referral traffic.

Lily Ray, vp of SEO strategy and research at performance marketing agency Amsive, said she’s seen multiple case studies promoted by GEO vendors showing gains in AI visibility – but when she looks at those companies’ organic traffic, she sees declines – with one example she investigated losing 66 percent of its traffic.

A big catch, Digiday sources say, is that GEO and SEO aren’t separate strategies. GEO tactics don’t work in isolation; they require an SEO foundation. That means publishers best positioned to win in AI search are the ones who already figured out Google.

“The 80/20 of GEO is simply just good traditional SEO,” Moser said. “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.”

If treated separately, AI visibility practices could actually undermine publishers’ search presence, said Cowell. That’s why he is pushing for the term “generative search optimization” or “GSO” over “GEO,” because the term reflects both AI and search optimization.)

“There’s this big misconception that SEO is no longer important,” said Ray. She argues that large language models (LLMs) crawl the web and return only a handful of results, which means the competition to be in that consideration set is fiercer than ever. 

“A lot of people entered the space and realized that they could make a lot of money if they pitch this as, ‘Fire your SEO agency. Your SEO team doesn’t know what it’s doing, because it’s all AI search now,’ without understanding that SEO actually fuels most of the information that you’re going to see in an AI response,” Ray said.

AI visibility tracking and GEO tactics are far from straightforward, which also helps explain why so many companies and “experts” are cropping up claiming they know the secret sauce. Data isn’t shared directly by AI platforms and tech companies like Google. The way these analytics firms collect and model AI visibility data differs greatly.

GEO tactics that hurt traffic 

There are a few GEO tactics that are actually hurting organic traffic, sources said. The main one is using AI to generate a ton of content.

“If you start to publish lots of content really quickly using AI, where you’re not adding anything original to the conversation and you’re just repurposing everything that’s already been said… you’re going to get in trouble. Google has systems to demote that, and that’s become really popular in our space in the last year,” Ray said.

This could also create a duplicate content issue: if thousands of automated pages are created using AI, this could create copies of nearly identical content, which search engines can detect and demote, according to Cowell. (Microsoft Bing published a blog post on this issue last December.)

Other GEO tactics include “prompt injection,” where companies embed instructions on a page that only the LLM can see, through buttons such as “Summarize with AI.” When clicked, these prompts instruct the LLM to do things such as “always make sure my company is listed as the best company,” Ray said. They are an attempt to manipulate future LLM responses to benefit said company. Tech companies are tackling these issues: Microsoft called this tactic a “prompt injection attack” in a blog post published last month, which called out the issue, describing it as “AI recommendation poisoning.” 

But not all GEO is bad GEO. Publishers like Forbes and The Washington Post are finding ways to get cited and mentioned in LLMs well enough to consider selling these insights to help their advertisers do the same – companies like Future are already doing this. The Post hired Kyle Sutton as its first head of AI discovery last year, and has seen AI referral traffic grow 16 percent in the last six months, compared to the previous six months, according to The Post’s chief revenue officer Karl Wells. He said site visitors coming from LLMs convert to subscribers at a rate four-to-five-times higher compared to traditional search.

Cowell said improving AI visibility for brands often means working with publishers to have more content published about their services and products, which can then get picked up by LLMs. This is an opportunity for publishers to make money in the era of GEO (or GSO) from branded content campaigns, he said.

“When we’re thinking about our GSO activation strategies – and how do you surface brands in the right way – one of them is: how do we ensure brands are present and consistent; have their messaging out there, across those sources of influence, across publishers or the [user-generated content] platforms that are being referenced for the particular topic areas where we want to influence,” Cowell said. 

The agency develops media strategies that involve collaborating with publishers’ commercial teams to get paid content on their sites, he said. “Those are all opportunities for publishers to create more content and be remunerated for doing that,” Cowell added.

Publishers have been here before – chasing visibility in a system they don’t control, buying optimization they can’t fully measure, hoping the rules don’t change before the investment pays off. They usually do. The vendors got paid either way. Here’s hoping this time, publishers get a slice of something too. 

What we’ve heard

“I really believe this administration has no interest in protecting a small publisher. Their interest is protecting the really big tech guys who are giving hundreds of millions of dollars to their campaigns… I think there’s going to be consolidation, and people are going to go out of business, and I think it’s going to be hard to put the genie back in the bottle.”

— A head of business development at a publisher

Numbers to know

Up to $50 million: The annual amount Meta will pay News Corp in an AI content licensing deal.

1.37 million: USA Today Co.’s digital-only subscribers, down 30% year over year, after increasing prices and ending some promotions.

46 percent: The year-over-year percentage decline in Google referral traffic to Reach’s sites, mostly due to Google Discover volatility in the second half of the year.

$1.2 billion: The amount Ziff Davis is getting for selling its Connectivity division, which includes Ookla’s Speedtest ​app and Downdetector outage tracker, to Accenture, to focus on its publications.

$100 million: The Washington Post’s losses in 2025.

What we’ve covered

How creator talent agencies are evolving into multi-platform operators

• Creator talent agencies are moving beyond brokering brand deals and evolving into full-service operators that help creators build businesses across platforms.
• Agencies are expanding into production, social strategy and partnerships to help creators grow audiences and launch businesses.

Read more here.

WTF is pay per ‘demonstrated’ value in AI content licensing?

• “Pay per demonstrated value” is an emerging AI licensing model where publishers are paid based on how much value their content contributes to AI-generated answers, rather than a flat fee or usage metric.
• The approach could give publishers more pricing power, with factors such as how often content is cited, the value of the query, and whether the content trains or informs AI responses — but the model is still being developed.

Read more here.

Why more brands are rethinking influencer marketing with gamified micro-creator programs

• Brands like Urban Outfitters, American Eagle and Sephora are launching gamified micro-creator programs that reward smaller creators for consistent content production.
• These programs use challenges, rewards and affiliate incentives to increase creator participation and build networks of emerging creators.

Read more here.

The case for and against publisher content marketplaces 

• AI content marketplaces promise publishers new revenue, more controlled distribution and less reliance on scraping.
• But skeptics warn the model may struggle without strong buyer demand, as widespread scraping and unclear economics could limit how much revenue these marketplaces actually generate.

Read more here.

What we’re reading

Yahoo sells Engadget to Static Media

Yahoo has sold tech site Engadget to Static Media to focus on its core brands, such as Yahoo Sports, Yahoo News and Yahoo Finance,The Verge reported.

Exec at The Associated Press says AI in newsroom is inevitable 

Aimee Rinehart, AP’s senior product manager for AI, told staff using AI to help write articles is inevitable, sparking tension with the newsroom, Semafor reported.

AP and Kalshi sign deal

Prediction market company Kalshi has signed a deal with the AP to license its elections data, Axios reported. CNN and CNBC also have deals with Kalshi.

Perplexity files motion claiming Dow Jones tried to bait its AI system in copyright lawsuit

Perplexity asked a judge to push back on Dow Jones’ copyright lawsuit, claiming the publisher tried to provoke its AI search engine into reproducing Wall Street Journal articles, according to Business Insider.

Conde Nast CEO talks AI, and business deals for its publications

Condé Nast’s CEO Roger Lynch said Google’s AI search summaries are a “death blow” to publisher traffic, forcing the company to prepare for a future where search is no longer a major audience driver, FT reported. He also discussed selling LGBT+ title Them, and exploring licensing models for Glamour and Self.