How to Measure Share of Voice Across Press, Social and AI

Share of voice is the proportion of all coverage in a category that belongs to one brand, usually expressed as a percentage of mentions against its competitors. It is only meaningful when the competitor set, sources, time window and counting rules are fixed and disclosed.

Share of voice is one of the few PR metrics that started out genuinely useful and got steadily less trustworthy as more vendors started reporting it, each with a slightly different definition, none of them disclosed clearly enough for a buyer to compare across tools.

The formula itself is simple. Getting a number that means something, and that survives being compared month over month, requires a few decisions made deliberately rather than left to whatever a vendor’s default happens to be.

The formula, and where it breaks

Share of voice is your brand’s mentions divided by total mentions across your brand and a defined set of competitors, in a defined time window, expressed as a percentage.

If your brand had 400 mentions last month, and three named competitors had 300, 250 and 250 respectively, total category mentions are 1,200, and your share of voice is 400 divided by 1,200, or roughly 33%.

The formula breaks the moment any of its inputs are left loose. “Mentions” needs a definition: does a syndicated wire story that ran in forty outlets count forty times or once. “Category” needs a boundary: does a competitor that operates in an adjacent but distinct market belong in the set. “Time window” needs to be fixed and comparable: a rolling thirty days measured differently each month produces noise that looks like a trend.

Choosing the competitor set honestly

The single decision that most changes a share-of-voice number is who is in the denominator, and it is also the decision most vulnerable to being quietly adjusted to produce a flattering result.

The honest approach is to fix the competitor set based on who a customer would actually compare you against, not who happens to generate convenient numbers. This usually means naming three to six real competitors, based on actual market overlap, and then holding that list fixed for at least a year regardless of whether the numbers it produces are flattering in a given quarter. Changing the competitor set the moment a number looks bad is the fastest way to make a share-of-voice trend meaningless to anyone who notices the list changed.

A related trap is including a competitor who is not really being compared by the same audience, which pads the denominator without reflecting anything a customer actually experiences. Naming an aspirational, much larger competitor in the set to make your own presence look proportionally smaller than customers actually perceive it is a common way this metric gets misused, usually unintentionally, by teams trying to seem appropriately humble in a report.

Why cross-channel share of voice is harder than it looks

The real complexity in modern share of voice is that a category’s conversation no longer lives in one channel. Press coverage, social conversation and now AI-generated answers each carry a different, only partially overlapping picture of who is winning the narrative, and combining them honestly requires a genuinely consistent methodology across all three, not three separate numbers stapled together.

This is where the single-vendor argument from adjacent measurement work applies directly. A share-of-voice number computed by one analysis engine, using the same mention-counting and deduplication rules across press, social and AI answers, can be honestly combined into one figure. A number stitched together from a press-monitoring vendor’s press count, a separate social tool’s social count, and a manual tally of AI mentions, cannot be combined without introducing exactly the kind of quiet methodological disagreement that makes the final number impossible to defend if anyone asks how it was built.

Including AI answers specifically is now worth doing deliberately rather than as an afterthought, because a growing share of category research happens through an assistant before a person ever visits a search engine or a review site. A share-of-voice metric that only covers press and social is measuring a shrinking fraction of where the actual comparison against competitors is happening.

Reading the trend, not the snapshot

A single month’s share of voice tells you almost nothing on its own. The number that matters is the trend: is your share moving up or down relative to your own history, and does that movement correspond to something identifiable, a major campaign, a competitor’s product launch, a crisis that consumed a rival’s coverage for a news cycle.

This is also where a fixed methodology earns its keep. A trend built on a consistent competitor set, source scope and mention definition shows real movement. A trend where any of those three things shifted between measurement periods shows methodology noise dressed up as a business signal, and reporting it as the latter is the fastest way for a comms team’s numbers to lose credibility the first time someone checks the history closely.

What to do with the number once you have it

Share of voice earns its place in a report when it is paired with an explanation, not presented alone. A share of voice increase driven by a genuine campaign is a different story than one driven by a competitor going quiet for reasons unrelated to you, and a report that distinguishes between those two cases is doing the actual job of communications measurement, which is explaining what happened, not just what the number was.

The metric is worth the effort to build honestly, because unlike reach or AVE, it answers a question that genuinely matters to a business: relative to the companies we actually compete with for attention, are we gaining or losing ground. That question survives scrutiny. The number only survives scrutiny alongside it if the methodology behind it does too.

Common questions

How do you calculate share of voice?

Divide your brand's mentions in a defined category and time window by total mentions across your brand and a named set of competitors in the same window, then multiply by 100. The number only means anything if the competitor set, time window and source scope are held constant every time it is measured.

What is a good share of voice percentage?

There is no universal good number. It depends entirely on how many competitors are in the set and how the category is defined. A share of voice of 25% against four evenly matched competitors is average. The same 25% against a single dominant rival is a strong result. Compare the number against your own history and category structure, not a generic benchmark.

Can you compare share of voice numbers from different tools?

Not reliably. Different tools use different source lists, different deduplication rules and different definitions of a mention, so two vendors measuring the same brand in the same month often produce different numbers. A share of voice trend is only meaningful measured consistently over time by the same methodology.

Should share of voice include social media and AI mentions, or just press?

It should include whichever channels actually carry your category's conversation, which for most brands today means press and social at minimum, with AI answers an increasingly important third channel. Excluding a channel where real conversation happens understates competitors who are strong there and overstates your own position.

How often should share of voice be measured?

Monthly is standard for a stable trend line, but the more important discipline is consistency: the same competitor set, the same source scope and the same time window definition every period, so month-to-month movement reflects a real change rather than a methodology shift.