More than half of PR teams now lean on AI to help measure and report results, and most say they trust the numbers it hands back. Muck Rack’s 2026 survey of 571 communications professionals found 51 percent report high AI measurement confidence in the metrics AI helps produce. The same survey found fewer than a third test that process on a regular schedule.
- 76 percent of PR professionals use generative AI daily, up from under 30 percent in 2023, according to Muck Rack’s 2026 State of AI in PR report.
- 54 percent use AI specifically for measurement and reporting, and 51 percent call themselves highly confident in the resulting metrics (Muck Rack, 2026).
- Only 31 percent test their AI measurement process on a regular schedule, and fewer than half use consistent prompts, the same survey found.
- Meltwater’s separate 2026 survey of more than 1,100 professionals found the metrics PR teams report most are still activity counts like placements and impressions, not business outcomes.
Confidence is easy to feel and hard to earn. That gap shows up fast once you put these two Muck Rack datasets side by side.
How many PR teams are actually using AI to measure their work?
A lot, and growing fast. Muck Rack’s 2026 State of AI in PR report puts generative AI use among PR professionals at 76 percent, up from under 30 percent in 2023. Editing is the top use case at 86 percent, followed by research at 76 percent and writing at 74 percent. Measurement sits lower, but it’s still where 54 percent of respondents now apply AI, per the separate Muck Rack 2026 State of PR Measurement report, which surveyed 571 professionals between September 22 and October 19, 2026.
That second report found 55 percent of PR pros have started measuring LLM visibility, nearly double the 28 percent who said so a year earlier, with 23 percent more planning to start. Add those up and nearly four in five PR teams now watch how brands show up inside AI answers, not just search results or press hits.
Why is AI measurement confidence rising faster than the rigor behind it?
This part doesn’t add up cleanly. Among the same group reporting 51 percent confidence in their AI-assisted metrics, 72 percent separately name accuracy as their top concern about AI tools, and 45 percent flag ethical risk. You don’t usually get both from the same room unless confidence rests on something other than verification.
It mostly does. Fewer than half of respondents use consistent prompts when generating AI measurement output, and only 31 percent test their process on a schedule, per Muck Rack’s data. A number nobody re-checks isn’t a measurement. It’s a guess dressed up before it left the building.
Does more AI use actually mean better measurement?
Not on its own. Meltwater’s 2026 State of PR report, run with We. Communications across more than 1,100 professionals, found activity-based metrics like media placements and reach still dominate what PR teams report, while stronger signals like share of voice and message pull-through rank lower. Twenty-one percent named measuring impact or ROI as a top barrier.
So the industry picked up a faster way to produce the same shallow numbers it already had. AI didn’t create the gap between activity and outcome. It just lets PR teams fill that gap faster, in language that sounds more precise than the count underneath actually is.
| What the survey measured | Share who say yes |
|---|---|
| Use generative AI in daily PR work | 76% |
| Use AI specifically for measurement or reporting | 54% |
| Highly confident in the metrics they report | 51% |
| Name accuracy as their top AI concern | 72% |
| Test their AI measurement process on a regular schedule | 31% |
Who owns the problem when the metric is wrong?
Often, nobody in particular. That’s the thread tying this together, and it’s what O’Dwyer’s PR News argued in November 2026: autofilling the likeliest answer from a corpus of past text isn’t the same as the judgment a PR pro builds from years of reading a journalist or an audience. Trust is earned and lost in moments, and AI output doesn’t carry responsibility the way a person signing off on a number does.
mGrowTech has tracked a version of this before in how PR measurement is adapting to AI-generated answers, and in the ownership gap around who tracks AI visibility. Both point at the same root issue: PR built new instruments faster than it named anyone to calibrate them.
What should a PR team actually do about this?
Five concrete moves close that gap.
- Pick one AI-assisted metric and test it monthly, instead of trusting the first output.
- Standardize the prompts used to generate any reported number, and keep a written version for reproducibility.
- Name one person who owns AI measurement accuracy, even as a part-time addition to an existing role.
- Pair every AI-assisted metric with one outcome measure, such as a tracked conversion.
- Treat a 51 percent confidence number as a starting point for an audit, not a finished answer.
Frequently asked questions
Is AI actually making PR measurement worse? Not exactly, but it makes unreliable numbers easier to produce at scale. Muck Rack’s 2026 data shows rising use and confidence without a matching rise in testing, so the risk sits in that gap, not in AI itself.
What is LLM visibility and why does it matter now? LLM visibility means tracking how and whether a brand shows up in answers from tools like ChatGPT or Google’s AI Overviews. Muck Rack found 55 percent of PR teams started measuring it in 2026, up from 28 percent the year before, because buyers increasingly read an AI summary before they read a press hit.
Should PR teams stop using AI for measurement? No. The data shows clear speed and quality gains from AI in research and reporting. The fix is adding verification, not removing the tool: consistent prompts, a regular testing schedule, and a named owner for accuracy.
Why do activity metrics like impressions still dominate PR reporting? Meltwater’s 2026 survey found they’re simply easier to produce than outcome metrics like share of voice, and AI tools made them even faster to generate, which reinforces the habit instead of breaking it.
These are self-reported vendor survey numbers, so treat the percentages as directional, not audited fact.
Sources: Muck Rack’s 2026 State of AI in PR report and 2026 State of PR Measurement report (571 respondents, surveyed September 22 to October 19, 2026); Meltwater and We. Communications’ 2026 State of PR report (1,100+ respondents, published January 2026); and O’Dwyer’s PR News commentary on AI and human judgment in PR agencies, November 2026.