Everything PRs need to know about AI Overviews

If you’ve noticed your online experience has become a little succinct lately, you’re not alone. Gmail, Slack, Facebook, even Zoom calls – they’re all getting neatly packaged into easily digestible recaps. 

Leading the charge is Google’s AI Overviews – those pithy summaries appearing at the top of our search engine pages with their streamlined blurbs, bolded sub-heads and bullet pointed lists. 

If we have entered “the next era of search”, one where “AI will do the work so you don’t have to” (as the Google video which launched AI Overviews in May 2024 put it), then it’s proving remarkably popular. 

Around 1.5 billion people use AI Overviews to answer queries each month (making Google’s traditional rankings of links seem outmoded). It stands to reason that time poor journalists doing their research will get their topline info from this narrative too.

But what if you’re a PR whose clients rely on traffic and ad revenue from their websites? Or if your client’s brand is continually being misrepresented in AI Overviews? Or if you’re a PR pro who’s been using SEO for years to boost brand visibility? AI summarisation is disrupting everything the PR industry knows about online search. Today, success increasingly depends upon inveigling coverage in these tiny, pithy chunks of text. 

“[Securing] visibility in Large Language Models (LLMs aka the architecture behind AI Overviews and ChatGPT) is becoming a KPI in its own right,” says Paul Stollery, co-founder of PR agency Hard Numbers. “The smart brands are systematically monitoring LLMs to understand which of their press releases has done well over the last six months.” 

AI summarisation is now on clients’ radars too, he adds. “It’s trickling up to boardrooms which are starting to get questions about it, while there’s also an internal pressure on in-house marketers and PRs to be seen to understand it… It’s a big difference from 12 months ago when it was very much ‘look at this new fancy thing.’”

AI Overviews isn’t the only summarisation tool: ChatGPT Search, Claude and agentic ‘answer machine’ Perplexity can also generate AI-driven summaries within seconds of searching. With Microsoft reportedly building similar features into Word, PowerPoint and Excel, and media publications looking set to follow (The Independent recently launched Bulletin, which produces condensed AI versions of its stories), getting smart with summarisation and our new ‘zero-click’ world is something PRs will have to embrace…

Boosting visibility

The content for AI Overviews and its summarisation arch-rivals needs to come from somewhere. If it relied on amateur blogs and Reddit posts, then it’d result in poor quality (and sometimes inaccurate) AI slop, turning users away. Instead, its LLMs are programmed to favour reputable sources – the same kind of publications PRs have been building relationships with for years. When Hard Numbers recently conducted research into the sources behind LLM answers on brand reputation, it found editorial media was responsible for 61% of answers. Social media provided less than 1% of these responses.

It’s not for nothing that OpenAI – the company behind ChatGPT – has been signing licensing agreements with many of the world’s leading media organisations, such as News Corp, the Guardian and the Financial Times. 

“The irony is media coverage appears to be used as source material more than any other content format in providing the ‘fuel’ for AI search answers, so gaining credible media coverage is more important than ever,” says AI PR/comms specialist (and CIPR trainer) Andrew Bruce Smith. “AI companies need quality content, even if they’re playing a fundamental part in destroying the business models of the very places that are designed to create this quality content… it’s a catch 22.”

Media websites might be receiving less clicks from AI Overviews – Mail Online’s clickthrough rate from AI Overviews is down 56.1% on desktop compared to organic Google search – but gaining earned media coverage for brands in authoritative outlets is as important than ever for PRs.  As Stollery says, “Everything is different but the same, in that brands need to appear in fresh and original content if they want to stand out.”

Best strategy

“You can’t get away with targeting just one or two publications – you have to try and get presence,” says Smith. 

“You could argue nothing has changed since Edward Bernays (the ‘godfather of public relations’) wrote Crystallizing Public Opinion over 100 years ago, when he said public relations was ‘channel-neutral’ and that it’s the PR’s goal to interpret the client to the audience and vice versa. But the number of ‘surfaces’ modern PR practitioners need to think about has increased. The coverage must be ‘contextually relevant’ too…”

Contextual relevance

PRs will want to ensure that their brand or client’s name appears alongside the contextual topics they want to be associated with (think ‘innovation’, ‘inclusivity’ or ‘agile’) rather than product category keywords (generic search terms used by customers such as ‘best shampoo?’). AI Overviews will learn to link your brand or client to these values. 

Optimising online content

With AI Overviews curtailing traffic to brand websites, clients (and their PRs) might feel as if their ‘owned content’ such as CEO-written blogs and brand websites are becoming obsolete. Indeed, the rapid rise of AI Overviews has seen clickthrough rates from Google’s search results fall by nearly 30% in the last year according to a recent study from search optimisation firm BrightEdge. 

However, even if fewer people are directly visiting brand websites, AI agents will still be dropping by. It’s not time to shut down the company website and sell its domain name just yet.

As Smith says, “In a world with AI Overviews, there’s no ranking, just an answer – either you’re in that answer or not. Your original content isn’t what people are going to see, but it will provide the fuel for the AI to decide ‘What will I give back to the person who’s performed the search?’ It’s a very different paradigm.”

“It’s being talked about that it’s still worth having a website, but you’d be optimising it for an AI agent, not human beings,” he adds. It could even upend the world of online content: “The idea we’d visit websites to consume content, which are structured for human beings to navigate” now becomes a “raw dump of content where the agents [AI] are encouraged to find the right bits and synthesise it to bring to the end-user. There’s a lot of hype around this, but you can see where it’s going…”

Meanwhile, Hard Numbers’ research found owned media drives 44% of answers on LLM to questions on brand reputation. Says Stollery, “People might not be clicking on content to go through to your website, but LLMs are still pulling information from it.”

His message to clients? “If you want to make a long-term investment in good quality media coverage and you’ve got four hours, the most value you’d get in terms of summarisation is optimising your online content, because there’s lots of low-hanging fruit.”

Better visibility

Structure your data
Structured data (i.e. organising it into a standardised format) helps AI understand content better. “LLMs love structured data, even more than search engines,” says Stollery. For brands, this may mean adding specific tags to HTML code.

Get familiar with FAQs
“Every consumer brand should have a well-structured FAQs section on their website,” says Stollery. “It’s the easiest thing any brand can do to influence those queries/answers, because it’s matching like-for-like how people are using these platforms.”

Make documentation available
“One thing we encourage B2B providers to do – especially if they’re an software as a service (Saas) platform – is have all their documentation available and optimised for visibility because if you think about what people are asking about your brand (i.e. “How does [name of Saas platform] integrate with other tools?”), the answer is usually in the documentation,” says Stollery. “Many platforms hide it away by gating this information or with logins.” 

A good example is US software firm Hubspot, which has its documentation available.

Answer the questions people will be asking
“Sometimes product releases can be self-serving, but you’ve got to ask the question, ‘What will the audience be using this for?’” says Stollery. “If you can place the answers to this type of questions in high-ranking applications, then you’ll show up in the summaries.”

Use bullet points and sub-heads (plus questions in the sub-heads too).
AI prefers content that’s easy to scan. “In the older days of SEO, you could put up a 2,000-word piece of content and Google’s algorithms would crawl and index it,” says Stollery. “Today, LLMs need to quickly find things. So, if you do have longer content, make sure it’s ‘crawlable’ [with sub-heads and bullet points] and structured in a way that you have a question-and-answer approach. If you’re asking questions in copy, make sure these questions are part of those sub-heads too.”

Is SEO dead?

Not yet, but there’s an inherent problem. Google/OpenAI won’t disclose the search queries people are typing, making it harder for SEO strategists to understand the motivation behind a user’s search, or which keywords are most valuable. 

“We’re seeing lots of SEO companies putting wine in old bottles by saying ‘Aah! Things have changed but you still need us’, but nobody knows what people are typing in because Google/ChatGPT/Perplexity aren’t telling us,” says Smith. “So, at best, people are trying to second-guess what they think people are typing in. Even if you get that bit right, it’s not what people are seeing because LLMs deliver a different result every time – even with the same search term. It’s not consistent.”

 “Search is dying, but very slowly,” says Stollery. “My advice to marketers/PRs is don’t stop investing in SEO or search but balance this legacy investment with AI content optimisation so you’re on the frontline of whatever comes next.”

Google’s AI Mode

In May, Google launched AI Mode in the US (it’s not yet available in the UK), which provides more conversational answers to users’ questions. Or, as Smith explains it, “If I want to follow up on an answer AI Overviews gives me, I must conduct a new search. With AI Mode, you’re chatting to the AI. It gives people what they want (i.e. the ‘answer’) but it keeps you on the search page – you’re not leaving Google!”

Lack of website traffic

“I think we’re in an exciting period in which everyone is starting to view this as important,” says Stollery. “The PR practitioners who’ll do best out of this are the ones who will seize this opportunity as education, or as a piece of thought leadership, or client relations. You need to have some frameworks and let your clients know you’re spending time thinking about this.”

Smith agrees. “We’re going to see more summarisation, such as using ChatGPT as a search engine. The best way to keep up with it? Visit Perplexity and ask it the same question!”

Perplexity

Perplexity is much-loved in the tech community for providing robust and (mostly) accurate answers to questions, as well as plenty of sources in a transparent way (there are also accusations it bypasses paywalls too). 

“Perplexity is dwarfed by Google in terms of the numbers using it, but for those that do use it, it’s like a Damascene conversion,” says Smith. Perplexity can also generate databases, create podcasts and apps based on your search terms. “You give it a brief and it says, ‘I’m not restricted to good old-fashioned search: I can write computer code that can actually do something useful for you.’”

He also wryly notes there’s a sense of smug one-upmanship among users. “Perplexity is over-indexed in the tech sector. It’s like the old joke: ‘How do you know if somebody’s gone to Cambridge University? A: They’ll tell you.’ It’s the same with Perplexity.”

How AI summarisation helps PRs

It can distil clients’ turgid PDF reports – the type of 200-page juggernaut that lands in PR inboxes, who are then tasked with finding newsworthy angles to promote it. 

However, Smith notes. “If you put a 500,000-word document into free ChatGPT, there’s no way it can pay its full attention to it: you’ll only get substandard answers. But if you fire up Google AI Studio, it’ll quite easily eat a half-million-word document for lunch and summarise [the most salient details].”

Another use is monitoring and evaluation: AI can help PRs track how and where their brand is referenced. “If you’re evaluating media coverage, why are human beings doing that? It’s so time-consuming,” says Smith. “Now, you can give your piece to AI and say, ‘Read that as if you are a member of that audience or stakeholder’. That qualitative analysis can be done at speed and at scale, for pennies.”

Repurposing content for different platforms is another way to fine-tune messaging. 

“Using AI to repurpose content for a range of different channels is relatively under-exploited,” says Smith. “PRs can get AI to craft press releases and summarise them for synthetically created audiences. You could create 10 versions and get AI to tell you which version is most likely to resonate.”

Find out more

Hear more on this from Andrew Bruce Smith at the CIPR's PR, Search and AI training course on 12 November 2025.
 

Christian Koch is an award-winning journalist and editor who has written for the Sunday Times, Guardian, Evening Standard, Metro, Director, Cosmopolitan, ShortList and Stylist.

You may also like

Anatomy of a PR campaign: ‘Do it for them’ by Boots Hearingcare
Is PR the future of wildlife conservation?
20 years on, how Make Poverty History became the first big charity campaign of the digital era.