Okay, I need to talk about something that's been quietly upending everything I thought I knew about search visibility. And no, it's not another Google algorithm update (though give it a week). It's about PR. Yes, PR. The discipline that most SEO professionals have spent two decades treating like a nice-to-have cousin who shows up at Christmas and talks about "brand awareness" while everyone else is discussing conversion rates.

Turns out that cousin just inherited the entire estate.

Here's the number that broke my brain: 94% of AI citations come from non-paid sources. Earned media—actual journalism, independent reviews, expert commentary in publications with editorial standards—accounts for 82% of all citations across ChatGPT, Claude, Gemini, and Perplexity. Paid placements? They barely register. Muck Rack analysed over a million cited links to reach these figures, and I've been staring at them for days like they're a magic eye poster that's about to reveal the meaning of life.

For anyone who's built their entire discovery strategy around paid search, sponsored listings, and carefully optimised product pages… well. This is the professional equivalent of training for a marathon and then finding out the race is now a swimming competition. (You're not wrong for running. You just also need to learn to swim. Urgently.)

I know that heading stings a bit, but hear me out.

When ChatGPT or Perplexity answers a question, it doesn't just rummage through its training data and wing it. It retrieves documents from an external index, evaluates them for authority and independence, and then builds a response citing the sources it trusts most. The operative word there is trusts. And it turns out AI systems have remarkably good taste in what counts as trustworthy.

A product review in a respected publication carries a trust signal that your own blog post—no matter how exquisitely keyword-optimised—simply cannot replicate. An expert quoted in an industry journal outranks your carefully crafted thought leadership piece. A feature in a news outlet with established editorial standards beats your sponsored content placement every single time.

This isn't the AI being mean to you personally. It's rational design. These systems live and die on answer quality. If Perplexity starts citing press releases as authoritative sources, users will notice the answers are rubbish and switch to a competitor. The incentive structure of AI search inherently favours editorial independence. (I spent an embarrassing amount of time being annoyed about this before I realised it actually makes perfect sense. My therapist would call this "processing.")

Kevin Indig's analysis of over 21,000 ChatGPT citations reinforces this from a structural angle. Across the topics studied, roughly 30 domains captured 67% of all citations in any given vertical. And the domains holding those seats aren't the ones with the cleanest technical SEO audits. They're the ones with genuine topical authority—sites that answer multiple related queries from a position of editorial credibility and depth.

Here's the kicker: citation reach—the number of distinct prompts a domain answers—matters more than raw citation volume. A single well-structured, authoritative page can outperform an entire content portfolio from a brand that publishes constantly but has never been independently validated. Which is either encouraging or devastating, depending on which side of that equation you're sitting on.

There's a temporal dimension to this that makes it more operationally demanding than just "get some press coverage and call it a day." (If only.)

Muck Rack's February 2026 analysis found that more than half of all AI citations come from content published within the previous 12 months. The highest citation rates occur within seven days of publication. Press release citations specifically grew fivefold between July and December 2025.

So AI systems aren't just preferring earned media—they're preferring recent earned media. A product launch feature from eighteen months ago is essentially invisible. That seasonal press hit you were so pleased about? Decaying as we speak.

This means a single episodic win—the kind of coverage that makes you screenshot the headline and send it to your mum—is not enough. AI systems reward ongoing editorial presence. A law firm that generates consistent commentary across legal publications, a SaaS company that maintains a steady cadence of genuinely newsworthy announcements, a local restaurant group that keeps showing up in food media—these will accumulate citation advantages that compound over time. A business that treats PR as a quarterly campaign will watch its citations evaporate within months.

(I keep thinking about this like compound interest, except instead of money it's editorial credibility, and instead of a savings account it's an AI's memory. The analogy breaks down if you push it, but the principle holds: consistency beats intensity.)

And even evergreen content isn't safe. Year-anchored titles, refreshed statistics, and current examples are structural markers that AI systems appear to use when evaluating relevance. That otherwise excellent "Ultimate Guide to [Your Industry]" that was last updated in 2024? At a measurable disadvantage. You need to update it. I'm sorry. I don't make the rules. (Well, nobody makes the rules. The AI sort of figures them out on its own, which is either fascinating or slightly unnerving depending on your relationship with ambiguity.)

Gartner's 2026 communications predictions are advising CMOs to double their PR and earned media budgets by 2027, explicitly recommending that the money be reallocated from paid channels. When Gartner tells you to move budget, the C-suite tends to listen—or at least schedule a meeting about listening.

Their 2025 CMO Spend Survey placed paid media at 30.6% of total marketing spend. If earned media budgets double, that money comes from somewhere. And the somewhere is increasingly obvious.

Paid search budgets face compression from two directions simultaneously: AI-mediated discovery is reducing the volume of traditional search queries that even reach the results page, and the AI systems replacing those queries preferentially cite earned media. Continuing to spend at current levels on a channel whose reach is declining while underinvesting in the channel AI systems actually trust is—and I'm trying to put this delicately—a strategic misallocation that gets more expensive every quarter.

This doesn't mean abandoning paid media. Paid channels still drive measurable, attributable revenue, particularly for conversion-oriented activity lower in the funnel. The shift is about discovery. The top of the funnel—where potential customers and AI agents alike are forming preferences—is migrating toward AI interfaces. And those interfaces trust earned media over everything else.

There's a conversion angle too. Semrush's analysis found that visitors arriving from AI search convert at 4.4 times the rate of traditional organic search visitors. (That figure is from digital marketing and SEO verticals specifically, so the exact multiplier will vary by industry, but the direction is clear.) AI search traffic is higher-intent because the discovery layer filters for relevance before the user ever reaches your site. They're not browsing. They've been sent.

Right. Practical bit. (I promise I'm not going to tell you to "leverage synergies" or "activate your earned media strategy." If I ever write that sentence unironically, someone please stage an intervention.)

Audit your AI citation footprint. Open ChatGPT, Perplexity, Gemini, and Claude. Ask the discovery-stage questions your potential customers would ask—not branded queries, but the upstream ones. "What should I look for in [your category]?" "How does [your brand] compare to [competitor]?" "What are the best [services] for [use case]?" Look at what gets cited. If your brand doesn't appear—and if the sites that do appear have no editorial coverage of you—that's the gap.

Identify the citation gatekeepers in your vertical. Indig's research shows citation dominance is concentrated: roughly 30 domains hold 67% of citation share in any given topic. Finding those domains in your space is the new keyword research. Which publications and editorial voices already hold authority? What kind of coverage earns citations versus what gets published but ignored by AI systems?

Build a cadence of genuinely newsworthy activity. This is where most businesses struggle, because PR requires something to say. Original research. Proprietary data. Expert commentary on industry trends. Partnerships. Supply chain innovations. A B2B software company publishing quarterly benchmark reports. An accounting firm releasing annual tax complexity analyses. A dental practice group sharing patient outcome data. These are the raw materials of earned media. Businesses that have nothing to say beyond "we exist and sell things" will find earned media structurally inaccessible. The editorial investment isn't just in the PR function—it's in creating the substance that makes coverage warranted.

Make sure your owned properties support and amplify earned coverage. When a journalist writes about your brand, the AI system that later cites that coverage will also check your own site to corroborate the claims. If your site lacks structured data, clear entity definitions, a maintained newsroom, and content reinforcing the same narratives being covered externally, the citation value of earned media is diminished. Owned and earned are not alternatives. They're mutually reinforcing layers of the same discovery architecture.

Start measuring citation share. How frequently does your brand appear in AI-generated responses relative to competitors? This sits alongside traditional SEO rankings and paid search impression share as a core visibility metric. Most teams haven't implemented tracking for this yet, which means the ones who start now will have a baseline from which to demonstrate ROI on earned media investment. The ones who don't will be making budget decisions blind.

The search industry spent twenty years building a discovery ecosystem around optimising for Google's algorithm. That discovery layer is now fragmenting across a dozen AI systems that each weigh signals differently. But the signal they all agree on is third-party editorial trust. Not backlinks. Not technical SEO scores. Editorial trust.

PR just got handed the playbook that used to belong to SEO. The smartest operators—whether you're running a SaaS platform, a professional services firm, a local business, or a publishing operation—will recognise this isn't a competition between PR and SEO, but a convergence. The brands that integrate earned media into their search strategy with the same seriousness, investment, and accountability they've applied to paid and organic channels will hold the discovery advantage in an AI-mediated market.

Those that continue treating PR as a brand exercise disconnected from search performance will find themselves structurally invisible in the places where their customers are increasingly looking.

And honestly? I never thought I'd write an article arguing that the PR people were right all along. But here we are. The data doesn't care about our professional allegiances. (My SEO friends are going to be furious with me. Worth it.)

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.

😍It was great, all I needed

🙂Good, not great

🙄It sucked