PR has an AI problem. But it’s not the one we think 

The PR industry is talking about AI constantly right now. AI-generated reporting. AI-powered insight. AI analysis layers. AI summaries. AI optimisation. 

But after months of research and conversations with PR leaders across agencies and in-house teams, I’ve become convinced that the industry’s biggest issue isn’t whether we’re using AI. It’s whether our data is ready for it. Because PR doesn’t just have a tech or measurement problem. It has a data structure problem. And AI is exposing it very quickly. 

Most PR coverage data still lives in spreadsheets, PDFs, screenshots, disconnected monitoring tools and manually maintained tagging systems. Different teams classify coverage differently. Different markets use different frameworks. In many organisations, the logic behind reporting still lives inside one person’s head. The result is that we’re trying to layer AI onto fundamentally inconsistent data.  

Our recent research with 160 senior PR and communications professionals across the UK and US found that 50% of PR leaders are not confident their data structure is reliable enough for AI analysis, while 45% are not yet using AI tools to analyse coverage data at all. 

That shouldn’t really surprise us. AI only works well when the inputs are consistent. Right now, PR’s inputs often aren’t. And this challenge runs far deeper than technology.  

Most PR reporting today is still retrospective. We spend huge amounts of time reporting what happened after campaigns finish. How much coverage did we get? Where did it land? How many impressions did it generate? 

Meanwhile, other marketing disciplines have evolved into real-time optimisation functions. Paid media teams optimise campaigns weekly. SEO teams benchmark performance continuously. Sales teams analyse pipelines daily. 

PR often still reports after the opportunity to influence the outcome has already passed. The issue is not effort. The issue is structure. 

Without structured data, campaigns cannot be reliably compared. Messaging trends become subjective. Cross-market analysis becomes inconsistent. And AI outputs become unreliable because the underlying classifications are unreliable. Every campaign effectively becomes its own island. 

What became very clear throughout the research is that this problem affects almost every role in the industry. 

Senior in-house communications leaders are increasingly expected to provide strategic intelligence, not just reporting. They need to explain which narratives are working, which markets are strongest and where investment should shift next. Yet fragmented data structures make those answers difficult to trust.   

At the same time, lean in-house teams are often carrying huge operational risk because reporting processes depend entirely on one individual maintaining a spreadsheet orundocumented workflow. More than half of PR leaders surveyed said they are not confident someone else could easily follow their measurement system if a key team member left tomorrow. 

Agencies face the same structural issues from another angle. Clients increasingly expect real-time strategic insight, but many teams are still manually cleaning spreadsheets, fixing inconsistent taxonomies and correcting duplicated classifications instead of analysing meaningful patterns. 

This is where I think the industry has standardised in the wrong order. For years, we focused on standardising KPIs, dashboards and reporting frameworks before standardising the operational layer underneath them: classification, taxonomy and data inputs. 

But KPIs should flex. Classification shouldn’t. That distinction matters. Because objectives will always change between campaigns. Outcomes should change too. A consumer PR campaign should not be measured in exactly the same way as a crisis campaign or a corporate reputation programme.  

But the structural layer underneath them should stay consistent enough to allow meaningful comparison. That is the missing layer in PR measurement. The direction I believe the industry now needs to move towards is consistent structural classification beneath flexible campaign outcomes. 

At CoverageBook, our research increasingly points towards four fixed structural drivers that remain consistent regardless of campaign type or geography: activity type, published format, sentiment and brand prominence.

That consistency is what unlocks comparability. And comparability is what unlocks strategic learning. 

Other industries already operate this way. SEO improves because rankings can be benchmarked over time. Paid media improves because campaign performance can be compared consistently. CRM systems improve because pipelines can be analysed across teams. 

PR, by contrast, still largely compares isolated campaigns using inconsistent systems. This is why I think AI is going to accelerate a divide inside the industry. The teams that solve their structure problem first will move ahead quickly. Not because they necessarily have better AI tools. But because they have better underlying data. And increasingly, clients and boards are going to ask harder questions about reliability, comparability and confidence in the numbers being presented to them.  

That’s why CoverageBook is now evolving beyond visibility and reporting towards a more structured data and insight layer designed specifically for modern PR teams. The opportunity is not heavier reporting or more complicated dashboards. It’s helping PR teams create cleaner, more structured, more AI-ready coverage data as part of the natural flow of PR work itself. 

Because the future of AI-powered PR will not belong to the teams with the biggest dashboards. It will belong to the teams with the cleanest, most comparable data.