Most public relations reporting falls apart the moment someone outside the communications team reads it. A chief financial officer sees a number labelled advertising value equivalency, asks how it was calculated, and the conversation ends badly. The problem is rarely the work itself. Coverage was earned, journalists were briefed, the story landed. The problem is that the media monitoring metrics chosen to describe that work were picked because they were easy to export, not because they described anything the business could act on. We run media monitoring for clients who sell in several markets, and the first thing we do on a new account is throw out roughly half the numbers they were previously reporting.
A defensible set of media monitoring metrics has a simple property: every number in it can be traced back to a decision someone would make differently if the number changed. That sounds obvious and almost nobody does it. Reach is reported because the tool provides it. Sentiment is reported because the tool colours it green. Neither figure, on its own, changes a media list, a spokesperson choice, a product message or a budget line. The metrics below are the ones we have found survive contact with a board meeting, and the reasoning matters more than the specific list, because your industry will shift the weightings.
Start With Share of Voice, Not Volume
Raw mention volume is a vanity number in almost every context. It rises when your category gets busy, when a competitor launches something, when an unrelated company shares your brand name, and when a single wire story is syndicated across two hundred low quality aggregators. Share of voice fixes part of this by placing your volume against a named competitive set, so a flat month during a noisy quarter reads correctly as a decline. Define the competitor set once, write it down, and do not quietly adjust it when results look poor. The most common way monitoring reports lose credibility is a competitor set that changes shape between quarters.
Then split share of voice by tier. A mention in a publication your buyers actually read is not interchangeable with a mention in a content farm. We usually work with three tiers defined by the client rather than by the tool: publications that influence purchase, publications that influence recruitment and investment, and everything else. The third tier is still counted, because a sudden spike there is often the earliest signal of a reputational problem, but it never enters the headline figure. Tiering is unglamorous manual work at setup and it is the single change that most improves how seriously monitoring output is taken.
The Media Monitoring Metrics Worth Reporting Monthly
Keep the monthly report short. A long report gets skimmed and a skimmed report gets misread. The set below fits on a single screen and each line implies an action if it moves in the wrong direction.
- Tier one share of voice against the named competitor set, with the raw mention count shown underneath so nobody suspects the percentage is hiding a collapse in volume.
- Message pull through: the share of coverage that contains at least one of the three or four messages you actually briefed, checked by a human on a sample rather than by keyword matching alone.
- Spokesperson and quote rate: how often your people are quoted rather than merely named, which is the clearest measure of whether you are a source or a subject.
- Referral sessions and assisted conversions from earned placements, pulled from analytics rather than from the monitoring tool, and reported as a range because attribution here is genuinely imprecise.
- Negative coverage in tier one only, with the time between publication and internal awareness, because response speed is the part of crisis handling you can improve in advance.
Notice what is absent. There is no estimated advertising value, no cumulative reach figure summed across outlets, and no single sentiment score for the whole month. All three combine unlike things into one number and none of them survives a direct question about methodology.
Sentiment Is a Sorting Tool, Not a Score
Automated sentiment classification is useful for triage and unreliable as a headline. It struggles with irony, with industry jargon, with quoted criticism inside a positive article, and with languages that the model was not primarily trained on, which matters enormously for teams operating across Turkish, Arabic, Russian and English at once. Use it to route articles into a queue for human review, and report sentiment only on the tier one subset that a person has actually read. A small honest number beats a large automated one.
Connecting Coverage to Commercial Outcomes
Earned media rarely converts in a straight line, and pretending otherwise invites a comparison with paid channels that public relations will always lose. What works better is treating coverage as an input to demand rather than a direct source of it. Watch branded search volume in the days after a major placement, watch direct traffic, watch whether sales conversations start warmer. Hold the reporting window open for several weeks, because trade coverage often surfaces in buying committees long after publication.
Finally, review the metric set once a year rather than continuously. Media monitoring metrics that change every quarter make trend analysis impossible, which is convenient for anyone who wants to avoid accountability and useless for everyone else. Agree the definitions, document them in a short methodology note attached to every report, and hold the line. The credibility of the whole programme comes from that note more than from any individual result inside it.
Keep reading: Media Monitoring · Digital PR