For the last few years, many in PR have argued that earned media influences far more than human readers. It shapes how machines understand brands too. That claim often sounded abstract…until now.
In a recent interview with Search Engine Land, Google’s Robby Stein confirmed that AI systems rely on public articles, lists, explainers and expert commentary as direct evidence when deciding what – and who – to recommend.
In other words, PR has always mattered to search, but AI makes its influence explicit. Brands that dismiss PR as a soft discipline are about to realise it governs whether they appear in AI answers at all.
Below is what Google’s comments mean for communications.
PR feeds the AI model’s reasoning, not just its results
The deeper point in Google’s comments is not simply that specific articles influence AI recommendations. It is that the overall reputation a brand builds through PR becomes the raw material AI systems use to form their understanding of that brand.
Models do not “look up” your marketing messages. They synthesise your reputation from whatever trustworthy, independent sources exist in the public domain.
That means earned media acts as the evidence base from which AI infers whether a brand is credible, safe and suitable for a particular user’s needs.
This is why PR cannot be reduced to a mechanical hunt for coverage. It must build a coherent and durable narrative across multiple touchpoints, e.g. national titles, trade media, expert commentary, long form interviews, rankings, reviews and industry guides.
Each of these contributes to the reputation profile that AI models assemble when judging a brand’s relevance.
For brands, this marks a shift from thinking about PR as episodic, individual campaigns, launches, stunts, to treating it as the long-term construction of an intelligible identity.
The rather simplistic question is no longer, “Did we get coverage this quarter?” but rather, “When an AI tries to explain who we are, what evidence will it find – and is that evidence coherent and authoritative?”
This broader, cumulative approach to reputation is where PR becomes indispensable. It builds the narrative scaffolding that AI relies on when piecing together brand understanding, not only for recommendations, but for everything from product comparisons to corporate history and category context.
From being ranked to being recommended
Search is no longer a list of blue links. It is a recommendation engine governed by reputation, authority and contextual cues. The classic SEO question – how do we get to number one? – is replaced by a more useful one:
Why should the model recommend us when someone asks a life-like, conversational question?
For PR, this is familiar territory. Awards, expert commentary, analyst citations, long-form features and trustworthy profiles all count as structured evidence that a brand is a safe recommendation. The more consistent the narrative, the clearer the model’s interpretation.
AI queries are longer, messier and closer to real decisions
Google notes that AI prompts are increasingly conversational and closer to real-world decision making. But this does not mean announcements are obsolete. A well-made announcement with substance, context and expert insight can be exactly the kind of material an AI needs to understand a brand’s competence, progress or category role.
What this shift does signal is the end of the thin, disposable digital-PR content that was never meaningful in the first place. AI is indifferent to stunts designed only to generate backlinks. It favours coverage that carries information, clarity and intent.
So PR must focus on material that actually says something. That may be an announcement, a data story, an expert view, a product update, or a practical explainer. The common thread is that it contributes to the evidence base from which AI forms its picture of a brand.
Rather than moving from “announcement-shaped” to “decision shaped”, the real requirement is a return to proper PR: credible information, clear narrative, informed commentary and messages that stand up to scrutiny. AI will elevate that material and quietly discard the fluff.
Owned and earned content should therefore be structured, consistent and genuinely useful. The model is looking for signals it can trust. Our job is to ensure those signals exist.
Local search becomes PR territory
Google’s demonstration of AI agents calling businesses, checking availability and synthesising reviews highlights a new reality: offline behaviour is now upstream of AI recommendations.
A Google Business Profile becomes a PR surface. So do reviews, FAQs, photos, menus and basic service information. Poor customer experience now leads directly to poorer AI visibility.
Local PR, once a niche discipline, becomes essential for brands that rely on service, footfall or geography.
Measurement will feel like the black box of prompts
We still have very limited insight into how AI systems make their recommendations. Tools such as Search Console, Trends and Ads can show what people are looking for, but they cannot show why an AI has chosen one brand over another.
AI queries are long, conversational and inconsistent, which makes their behaviour harder to track. And crucially, there is not a single tool on the market that offers a reliable or complete prompt library, so we cannot reverse-engineer the model’s reasoning by looking at “what people asked”.
For now, much of AI decision-making remains a black box.
We do not yet have a clear line of sight into how AI systems weigh the evidence they draw from earned, shared or owned media. The prompts themselves are invisible, the model’s internal reasoning is inaccessible, and there is no dependable tool that can tell us what people are asking in AI mode.
Because of that, the route from prompt to AI answer to human action is not traceable in the way traditional search behaviour once was.
A person may ask an AI for the “best X”, read one of the linked sources, and convert via a completely different channel. None of that journey is captured cleanly.
This means we cannot rely on old-style attribution chains. Instead, PR needs to work with strategic indicators that reflect how the model is likely to interpret a brand. These include:
- The volume and quality of authoritative press coverage and mentions
- How often experts and trusted outlets cite the brand
- The consistency of the narrative across independent sources
- Qualitative testing of AI answers (“Ask Gemini/ChatGPT/Claude what it recommends and why”)
These signals offer the closest picture of how AI is building its understanding of a brand, even if the underlying mechanics remain invisible.
This aligns with PRAO’s research on AI-driven visibility, which shows that more than 75 per cent of AI brand references originate from earned, shared and owned PR inputs .
PR, SEO and product now overlap
Google’s comments highlight that AI optimisation and SEO draw on many of the same signals, but this should not be misread as PR becoming a branch of SEO.
If anything, it reinforces a long-standing truth: PR has driven authority, reputation and external validation for years, and SEO has historically ridden on the back of that work.
AI simply makes the relationship more visible. SEOs can refine site structure, schema and technical crawl paths, but only PR creates the independent, high-trust signals that both search engines and AI models depend on when judging which brands are credible and safe to recommend.
What changes now is not the hierarchy but the coordination. AI forces PR, SEO and product teams to plan together because they are influencing the same reputational evidence base.
Brands should therefore establish a standing agenda item on AI visibility and recommendation readiness, ensuring that narrative, authority and technical performance align… on PR’s terms, not as an afterthought.
What PR teams should deliver next
Based on Google’s remarks, the practical deliverables for modern PR include:
- Coverage in authoritative news articles. They mention “best X” lists and guides, but the reality is coverage needs to be deeper and richer. Think about what is best for your brand reputation and start from there
- Explainers that mirror natural question patterns
- Regular expert commentary to refresh the knowledge graph
- Structured owned content designed for retrieval
- Audits of how AI models describe and recommend the brand
- Closer cooperation with customer-service teams, ensuring reviews and basic information support, not undermine, AI visibility.
The bottom line
PR has never sat outside search. It has always shaped how systems understand brands. Google has now confirmed what many of us have argued for years: earned media is not decoration. It is the scaffolding AI uses to build answers.
If a brand has an AI strategy that does not include PR, then it does not have an AI strategy. It has a deck.
