Digiday Publishing Summit:

Hear from execs at The New York Times, Thomson Reuters, Trusted Media Brands and many others

SECURE YOUR SEAT

Media Buying Briefing: Havas joins race to find answers to clients’ zero-click search queries

This Media Buying Briefing covers the latest in agency news and media buying for Digiday+ members and is distributed over email every Monday at 10 a.m. ET. More from the series →

As users adopt AI search tools like Google’s AI Overviews or ChatGPT in lieu of traditional search engines, clients are worried about what exactly generative AI search tools are saying about their brands and whether that can be measured and ultimately, influenced.

French holding company Havas is the latest to start offering some answers to their questions. It’s developed a tool called Brand Insights AI which can give advertisers a real-time ranking of their performance in search results generated by ChatGPT, Perplexity, Grok, Google, DeepSeek and Anthropic’s Claude model. The application has been used to ferret out key sources behind an AI response, and to provide recommendations for content that could indirectly exert more influence over zero-click results.

“The way people discover brands, information and products is changing rapidly… Brands [are] grappling with understanding the visibility of their brands and their products across the breadth of those platforms,” said Paul Bland, chief digital officer, Havas Media Network U.K. Bland said the tool was developed by Havas’ U.K. team and has been deployed since July.

Lower organic search click-through rates (CTR) and lower rates of traffic to websites overall are becoming the new normal for clients; 44% of advertisers saw flat or declining traffic since AI Overviews were launched on Google last year, according to a survey of 600 marketers published by performance agency NP Digital. Both marketers and their bosses are rattled.

Related Insights

“We’re pretty much talking to all of our clients about this stuff… they’re getting pressure from their board to be more visible in AI search,” said Nikki Lam, vp of SEO at NP Digital.

There is pressure to understand what is resonating among search results. At Wpromote, execs told Digiday they expect web traffic from AI search results to overtake “traditional” search traffic by 2027, based on projections built on data from Polaris, the agency’s proprietary measurement platform. (SEO firm Semrush, by contrast, has projected 2028 as the crossover point).

“Organic traffic overall is down because of this shift in user behavior to leverage more of these AI platforms to do almost the entirety of the consumer journey,” said Rachel Klein, svp of owned and earned media at Wpromote.

It’s as yet unclear how such a major change in user search behavior might affect brands’ approach to paid search. “There’s not enough data to really conclude anything,” Kenneth Yau, paid search managing partner at Dentsu told Digiday.

What is clear is that understanding those changing search habits has become a business priority for many clients. But Havas is not the first agency to push its own tool; several marketing services groups have already begun using their own proprietary solutions to address clients’ SEO concerns including indies like Kepler, Wpromote, R/GA, and Brandtech’s Jellyfish. Interpublic uses its Interact software platform, while Dentsu has developed its own tool called Deibu.

Meanwhile, a flush of start-up firms hoping to provide third-party applications that solve for AI search visibility have emerged, from VC-backed Profound to WriteSonic, Acme.bot and British start-up Azoma. Similarweb and DemandSphere also offer their own tools.

“We help our customers understand how they are perceived by AI engines,” explained Max Sinclair, Azoma’s co-founder. The company’s tool aims to provide a window on how LLMs like Gemini and Llama present information about brands, but also encompasses the search engines used in e-commerce environments, such as Amazon’s Rufus and Walmart’s Sparky chatbots. 

Michael Sondak, svp, and head of search for Omnicom Media Group North America, said the group’s global SEO teams had audited 30 different measurement tools so far this year — testing both software and the workflows required to get the most out of it. Sondak didn’t say which tools OMG had tested. 

Given the speed of change within the search discipline (within the last month alone, Google launched its AI Mode in the U.K. and OpenAI released ChatGPT 5.0), it hadn’t yet settled on a single approach, he added. “We are actively testing many tools today, as well as in-house solutions,” Sondak said.

Search practitioners aren’t just taking measurements. They’re working to remedy client AI search performance and in the absence of paid ad products running against zero-click search results (bar Google’s limited allowance in some markets) that means turning to traditional SEO techniques as well as the content marketing playbook: refreshed page content, schema tune-ups and architectural alterations that better lean into the kinds of material AI search applications draw upon and prioritize. 

(Google’s own executives have signalled that standard issue SEO tactics should be enough to get websites noticed by its AI crawlers).

The technical details vary but most of the applications work using the same basic ideas. First, they ask a given LLM about a brand hundreds or thousands of times to generate a large sample of responses. Next, the tools isolate common themes across those responses — as well as which sources the models are relying on — to better understand how an LLM is representing an advertiser to regular users, how it’s showing up compared to rivals, and how it might improve its position.

While this methodology can give clients vital clues into how they’re being represented in AI search, it’s a case of looking through a glass darkly. The only people who can give an authoritative reading on the inner workings of Gemini, Claude and Llama are the engineers behind them — outside applications provide a kind of brute-force guesswork that has to be repeated through updates and patches, and which comes with a caveat. 

“They’re directional at best,” said Daniel Toplitt, evp, search and digital experience at IPG’s Kinesso media agency. “You’re essentially making up prompts that are intended to simulate what real users are doing.”

“You’re sort of spamming these platforms to get results that aren’t necessarily representative of users. It never will be a real view, so we do need to be mindful of that,” noted Rachael Murdoch, managing partner, SEO, Dentsu UK

That aside, most clients are concerned enough to press ahead and increase their investment in SEO or GEO work. The opportunity cost of doing nothing is too high, argued Havas’ Bland. “It’s critical to understand where your gaps are so you can start to understand where your opportunities are and where to influence,” he said. “The risk of not being in this space is bigger.”

Color by numbers

In case you needed evidence that mobile app usage is altering the DNA of media consumption, vertical video continues to gain ground over the horizontal shot. 

A survey of 550 sports organizations and broadcasters by WSC Sports, a firm that provides content automation tech to sports rights holders, revealed that vertical video creative output shot up in the last year, as media owners cottoned on to viewers’ altered preferences.

The bottom line? Expect to see a lot more made-for TikTok snippet from the Premier League following Saturday’s season kick-off.

  • Vertical video production rose 87.5% year-over-year. 63% of content on WSC’s platform was vertical — a 18.6% rise.
  • Tennis saw the most vertical video production, with a 58.7% increase in videos per match; soccer saw a 29.5% leap; basketball increased vertical video output by 26.2%.
  • The average number of videos made by a sports media owner rose to 89, up 33.5% from 2024.

Takeoff & landing

  • The prospect of a merger between S4 Capital and MSQ Partners’ had M&A observers all aflutter this week. A deal might not materialize, especially after MSQ issued a note to the market quashing the idea, but the episode has served to elevate the group’s profile. I wrote about the group’s past and present, and its telling parallels with S4.
  • Dentsu’s second quarter results came in this week, bringing a 0.2% revenue decline across the board; its Japanese business an invariable standout from the broader network, bringing 5.3% organic growth to the table. In response, CEO Hiroshi Igarashi announced another new strategic re-ordering program that will see it lay off 8% of its staff worldwide.
  • Meanwhile, Cheil has brought together its North American agency businesses (including Barbarian, newly-merged Attention Arc and Iris) under a single banner dubbed the Cheil Agency Network. And IPG Mediabrands’ small creative production arm Traverse 32 has been moved into McCann.
  • Personnel moves: Accenture Song has a new boss in the USA. Following Ndidi Oteh’s nod for the global CEO job, Ami Palan is taking over as Americas lead. And WPP tapped Mindshare global chief solutions officer Tom Johnson for a role as the group’s evp of consulting growth and strategy.

Direct quote

“The plan is to move away from selling people’s time, services and their own inventory and move to selling platform-based solutions.”

– A former WPP Media exec’s summary of the media giant’s strategy for a turnaround, in Thursday’s feature on the firm’s future.

Speed reading

  • Michael Bürgi took a look at some of the investments media agencies are making in agentic tools and talent.
  • I reviewed some of the paid media strategies favored by brands looking to make the most of the back-to-school period, including JanSport, Nuuly and the plan behind the pushback at American Eagle.
  • Kimeko McCoy spoke anonymously to an agency exec about the conversations going on behind closed doors about how closely brands should stick to AI regulations when using the tech in their campaigns.
  • And Seb Joseph and Ronan Shields explored the AI and data-focused strategy that WPP Media’s execs hope can help it recover from recent missteps.

More in Media Buying

The Rundown: Ad tech’s Q2 report card

Fortunes are fluctuating, despite earnings increases across the board… time for consolidation?

From Dentsu to Apollo Partners, the buildout of AI tools and talent at agencies is picking up pace

The adoption of AI tools like agentic offerings and smarts in new hires to power more of media agency workings continues to pick up pace.

The Trade Desk holds its own against Amazon’s growing, but still distant threat

The Amazon threat to The Trade Desk is inevitable — just not immediate.