Public relations has measured its worth the same way for forty years. Share of voice. Impressions. Coverage volume. Tier-one placements. For most of that time, those metrics mapped reasonably well to how people actually discovered brands — through the media they consumed.
That mapping is breaking in real time.
Gartner projects a 25% decline in traditional search engine volume by 2026 as users shift to AI chatbots and virtual agents. Pew Research found that 58% of U.S. adults encountered at least one Google search with an AI summary in March 2025 — and when those summaries appeared, click-through to traditional links dropped from 15% to 8%. Only 1% of users clicked a source cited inside the summary itself. ChatGPT has crossed hundreds of millions of weekly active users.
Translation: audiences aren’t reading the coverage anymore. They are reading what the machine synthesized from the coverage — and most of the time, they never click through at all.
This is the citation economy. The metric that matters is no longer how many places your brand was mentioned. It is how often your brand, your executives, and your point of view get cited inside the AI-generated answer when a customer asks ChatGPT, Perplexity, Gemini, or Google AI Overviews a category-defining question. Call it Share of AI Voice. Share of Model. Whatever label we land on, every CMO and head of communications should be asking two questions by end of quarter: when someone asks ChatGPT “best [our category],” how often do we show up? And when we do, are we represented accurately — or is the AI quoting a three-year-old article we wish would disappear?
Most brands don’t know. Most agencies aren’t measuring it. That gap is the single biggest opportunity in PR right now, and it’s closing fast.
Here is what the shift actually looks like in practice.
Earned media becomes training data. In 5WPR’s GEO practice, one observation shapes almost everything we recommend: the majority of citations in AI-generated answers originate from earned and semi-earned environments — media coverage, expert quotes, executive commentary, and community platforms. These are the exact inputs PR teams have always owned. What changes is that the machine is now reading them too.
The distinction between SEO and GEO matters. SEO is about how machines rank content. GEO is about how machines interpret trust. The work involved looks similar on the surface and is fundamentally different underneath. We approach GEO at 5W through four pillars: semantic content optimization, structured data deployment, authoritative third-party source development, and always-on monitoring of how AI systems describe our clients.
Wikipedia becomes non-negotiable. Pew’s data confirms that Wikipedia, YouTube, and Reddit are the most commonly cited sources in both AI summaries and standard search results. Most CEOs and mid-market brands either don’t have a Wikipedia page or have one that is out of date. That is an unforced error.
Measurement becomes near-real-time. In the old model, you waited a quarter to see whether a campaign moved perception. GEO measurement can now show you within days whether your narrative is taking hold inside the platforms where purchase decisions are increasingly being shaped — which spokespeople are getting pulled into answers, which narratives are resonating, and which competitor stories are displacing yours.
The communications leaders who will look smart in eighteen months are the ones already having the uncomfortable conversation with their CFO: our impression counts are going up and our actual influence is going down, because the audience is behind the AI.
None of this makes traditional PR obsolete. It makes it more important. The inputs that train the models are the inputs PR teams have always controlled. What changes is the measurement layer, the narrative discipline, and the speed at which we need to move.
Share of voice told us who was being talked about. Share of AI voice tells us who is being recommended. Only one of those drives revenue in 2026.
