Is PR Dead? What Died, and What Didn't

No. The PR playbook that measured itself in clips, impressions and share of voice is dying, but the work of earning a credible outside source's judgment isn't. That work now has a second reader, the AI engines buyers ask first, and placement is the raw material those engines cite, which is why I call what PR has become Machine Relations.

"Is PR dead?" is really two questions stuck together. One is about a way of working. The other is about a discipline. They have different answers.

What actually died

The part that died is the scoreboard.

For a long time, PR was judged by what you could count in a monthly report: how many clips, how many impressions, how much share of voice against competitors. The job was considered done once the article went live. Agencies billed for activity, clients paid for it, and both sides agreed to treat the report as the result.

That model assumed the only reader was a person who might stumble across the story that week. When buyers start their research by asking ChatGPT, Gemini, Claude or Perplexity, a clip report tells you almost nothing about whether the coverage did its job. A placement nobody retrieves and an impression count nobody can connect to a buying decision are not worth much.

So if "PR" means paying for activity and reading a report, I understand why people say it's dead. That version deserves to go.

What survived

The discipline underneath survived, and that's the part that matters.

PR at its best was never about the clip. It was about getting a credible third party to look at a company and say something specific about it in public. A journalist or specialist publication applies scrutiny a company can't apply to itself. A buyer reads that and trusts it more than a self-description.

That hasn't changed. What changed is who reads it. As I wrote in why I coined Machine Relations, AI search adds another reader to that process. An answer engine can retrieve the coverage and use it to describe, compare or recommend a company. The judgment a credible source puts on the record now works twice: once on the person reading it, and again on the machine answering that person's question later.

Why AI made earned media matter more

The best data I've seen on this comes from Muck Rack's What Is AI Reading? study. The May 2026 edition analyzed more than 25 million links cited by ChatGPT, Claude and Gemini across 17 industries. It found that 84% of citations came from earned media, and paid and advertorial content made up just 0.3%. Earned media has ranged from 82% to 89% across all three editions since July 2025.

That number needs to be read carefully. Muck Rack's "earned" is broad. It includes academic research, government sources, encyclopedic sites and third-party corporate content, not only press coverage. The PR-relevant slice is journalism, which accounted for 27% of all cited links. That's still a large share of what these systems read, and it's a share you can't buy your way into.

Two other findings matter for anyone paying for PR. First, recency: more than half of journalism citations came from articles published within the past 12 months, with citations dropping sharply after the first six months. Coverage has a shelf life, so one launch-day burst won't carry a company for long. Second, the question shapes the source. Industry-trend questions pulled in journalism at more than twice the rate of how-to questions.

The full report and methodology are worth reading. The mechanism behind it is simple. When an engine needs to say something credible about a company, it reaches for sources that aren't the company.

How PR gets judged now

If the scoreboard died, something has to replace it. I think the test is simple: not whether the article ran, but whether the machines repeat it.

That changes what a good placement looks like. It needs to say something specific about what the company does and why a claim deserves belief. It needs to be in a source the engines actually read. And it needs to show up when a buyer asks about the category, not just when someone searches the company's name.

Share of voice measured how loud you were next to competitors. Share of citation measures whether you're one of the sources the answer is built from. I've written separately about why most earned-media ROI tools miss this, so I won't build a metrics framework here.

Why the debate keeps coming back

You can find evidence for either side this month. PRSA's PRsay ran a piece headlined with Muck Rack's view that AI citations are a "renaissance moment for PR and communications." The next day, Axios reported Indeed Hiring Lab data showing U.S. communications and PR job postings down about 31% since July 2023, compared with a 26% decline across all postings.

Both are true at once. The tasks are changing, and some of the old ones are disappearing. The reason to do the core work got stronger. People who look at one half of that picture say PR is dead. People who look at the other half say it's never mattered more.

Is Machine Relations just PR with a new name?

No. Machine Relations is PR evolved, not a rebrand. The raw material is the same: earned coverage from credible sources. What's new is the reader, and the standard that comes with it. I coined the term in 2024 because the discipline needed a name that covered both the authority communications earns and the discovery AI increasingly mediates.

If you want the full definition, start with What Is Machine Relations?

What I would do if I were paying for PR today

I'd keep paying for placements and stop paying for activity reports. Then I'd ask ChatGPT, Gemini and Perplexity the questions my buyers actually ask about my category, and look at which sources come back. If my coverage isn't in those answers, that's the gap to work on. It isn't the budget line to cut.

FAQ

Is PR still worth it in 2026?

Yes, if it earns coverage in sources that buyers and AI engines trust. No, if what you're really paying for is a monthly activity report.

Does AI replace PR?

No. AI engines lean heavily on earned sources to build their answers. AI changes who reads the coverage, not whether coverage matters.

Are press releases dead?

Not dead, but limited. Muck Rack found press releases appear almost exclusively in answers to industry-trend questions, at 3.5 times the rate seen in best-of queries. A release is a source document an engine can read. It's not the same as an independent publication choosing to write about you. More in Do press releases and wire services get cited by AI?

What replaced share of voice?

Share of citation: whether credible sources about you are among the sources cited when a buyer asks AI about your category.

Is PR dead for startups?

No. A startup has fewer outside sources describing it, so each credible one carries more weight. The same test applies: does the coverage show up when buyers ask? More on PR and AI search is collected in this topic.