Measuring Public Relations: What the Numbers Actually Tell You

Measuring public relations has been the discipline's unresolved problem for a hundred years. PR produces influence, and influence resists a clean number. The industry has cycled through impressions, advertising value equivalent (AVE), share of voice, sentiment, and reach — and each metric has been picked apart by the people who have to defend it to a CFO.

This is the EPR reference on how PR measurement actually works, why the standard metrics get abused, and what a defensible measurement program looks like.

The Impressions Problem

Impressions is the metric that has done the most damage to PR measurement credibility. The unit was originally intended to capture the size of the audience a piece of coverage reached. It has become the size of the audience a piece of coverage could theoretically have reached — which is a different number entirely.

A story on a page of a website that received 100 million visits that month is reported as 100 million impressions, regardless of whether anyone read the story. A wire release picked up by a syndication network of 500 outlets is reported as the combined unique monthly visitors of all 500 outlets. Vendors compete on which methodology produces the biggest number. CFOs learn to discount the numbers by an order of magnitude, and then by another. The unit collapses under its own weight.

Impressions are not useless. They are useful as a directional signal — did this campaign generate more coverage than the last one — and useless as a commercial metric.

Advertising Value Equivalent (AVE)

AVE assigns a dollar figure to a piece of earned coverage by pricing it as if it were a paid advertisement of the same size in the same publication. The metric is popular because it produces a big number and translates PR output into a language finance teams recognize.

It is also indefensible on its own terms. Earned coverage and paid advertising are different products. Editorial context carries credibility a paid ad does not, and it also carries editorial framing the brand does not control. Multiplying AVE by two, three, or five to account for "credibility" — a common vendor practice — makes the number less defensible, not more. The International Association for the Measurement and Evaluation of Communication (AMEC) formally rejected AVE as a measurement standard in the 2010 Barcelona Principles and reaffirmed the rejection in subsequent updates.

AVE persists because it is easy to calculate and hard to replace. Any program that uses it should treat it as a rough benchmarking tool, not a return-on-investment metric.

Share of Voice

Share of voice measures the brand's proportion of category coverage against direct competitors — how much of the conversation the brand owns. The metric holds up better than impressions or AVE because it is comparative rather than absolute.

The weaknesses are two. First, share of voice is only as good as the query set that defines the category. A poorly bounded category produces a share number that looks good and describes nothing. Second, share of voice is a volume metric. A brand can win share of voice on unfavorable coverage. Any share-of-voice program worth running is joined to sentiment classification.

Sentiment

Sentiment classification — positive, neutral, negative — is the standard reputation-monitoring layer on top of a coverage program. The category has matured. Human-coded sentiment produces reliable results at small scale. Automated sentiment classification has improved but still misclassifies sarcasm, industry-specific framing, and context-dependent language.

The practical discipline is to use automated sentiment for volume and human review for tier-one coverage. Any single sentiment score reported without a confidence interval is a report the CEO should not act on.

Reach and Engagement

Reach measures how far a piece of coverage traveled. Engagement measures what people did with it — clicks, shares, comments, time on page. Digital measurement has made both metrics more accurate than the impressions era allowed.

Reach and engagement are strongest when tied to a specific asset with a trackable URL — an owned press page, a landing page, a piece of research the brand published. They are weakest when tied to earned media the brand does not control, because the tracking infrastructure lives on someone else's property.

Building a Defensible Program

A defensible PR measurement program operates on four practices.

Define the objective before the metric. Awareness, consideration, reputation, crisis containment, recruiting, and pipeline are different objectives and require different measurement stacks. A program built around a single metric applied to every objective produces numbers that mislead leadership.

Report tier and quality, not just volume. A New York Times, Wall Street Journal, Bloomberg, Reuters, or Financial Times placement is a different commercial event than a trade-publication mention. Any measurement program that treats them as equivalent has already lost credibility.

Use comparative benchmarks. Numbers reported in isolation are noise. Numbers reported against the prior period, against direct competitors, or against a pre-campaign baseline are signal.

Tie the measurement to a commercial outcome. The mature version of PR measurement joins coverage data to pipeline movement, category conversion, brand-tracker movement, or reputation-index movement. The join is what turns measurement into a business report the CEO reads instead of a communications report the head of comms reads.

Related EPR Coverage

• Public Relations Pillar
• Reputation Management
• Crisis Communications
• PR Firms Directory