What Is Share of Voice?
Share of voice is a visibility metric that expresses your presence across a chosen set of keywords as a percentage of the total visibility available in that set.
Share of voice is a visibility metric that expresses your presence across a chosen set of keywords as a percentage of the total visibility available in that set. It is not a traffic figure. It counts estimated exposure across a list somebody selected, so two tools reporting on the same website can return different percentages without either being wrong.
Why the number gets quoted so often
Share of voice survives in reporting because it does something no other metric does: it puts you and your competitors on the same chart. Rankings move in ways that are hard to summarise. Traffic depends on seasonality, brand spend, and whatever else was happening that month. A single percentage looks like a scoreboard, and boards like scoreboards.
The trouble starts when it becomes the headline. A rising percentage can sit alongside falling revenue, because visibility across a keyword list and demand for what you sell are different things. Teams then defend the chart instead of the business. If a monthly report leads with share of voice and buries the commercial outcome, ask for the version that shows what the work actually produced, then judge both together.
How share of voice is calculated
There is no single formula, and that is the first thing to understand. Every tool builds its own version from the same three ingredients, and the choices made at each step decide the answer before any measurement happens.
- Someone defines the keyword set. A list is chosen, whether by a tool’s default, a competitor comparison, or a human deciding which terms count. Nothing about this step is objective.
- Each keyword gets a weight. Search volume is the usual weighting, so a handful of high-volume head terms can dominate a set of hundreds.
- Positions are converted to visibility. A click-through curve turns each ranking into an estimated share, because position one and position eight are not worth the same exposure.
- Weighted scores are summed. Your visibility across the whole set is added up, then divided by the total available across all tracked domains.
- The result is expressed as a percentage. One number, presented as though it described a market rather than a spreadsheet.
Step one carries the most weight and gets the least scrutiny. A set of 50 commercial terms and a set of 5,000 informational ones describe the same website and produce percentages that share nothing but a name. Tools also differ in how they estimate rankings, how often they refresh, and which devices and locations they sample, and none of that is visible in the final figure.
Two details matter more than the formula. Click-through curves are estimates, not measurements, and they were modelled on results pages that keep changing shape. Brand terms, if included, inflate your own score and nobody else’s. Neither problem appears on the chart, which is why the percentage should sit beside the sessions search actually delivered rather than replacing them.
What changes the number without changing your site
| Change made to the measurement | Effect on your percentage |
|---|---|
| Brand keywords added to the set | Rises sharply, since you rank first for your own name |
| A dominant competitor removed from the comparison | Rises, because the denominator shrank |
| Long-tail terms swapped for head terms | Usually falls, as volume weighting favours the biggest queries |
| Tracking location narrowed to one city | Moves either way, and is no longer comparable to last month |
| Tool updates its click-through model | Every historic figure is restated, without your site changing |
The blunt version
Share of voice is defined by whoever picks the keyword set, and in most engagements the agency picking it is also the party being graded by it. That is a conflict, and it is structural rather than dishonest. Nobody has to falsify anything. They only have to choose the list.
Work it through. Say the tracked set holds 200 keywords, 60 of them containing your brand name. You rank first for all 60, and those terms carry heavy volume, so they contribute most of the weighted total. The reported figure looks strong. Remove the brand terms and measure only the 140 non-brand queries, and the same site in the same month reports a fraction of that. Nothing about performance changed. The measurement boundary did.
So make the set the contract. Agree the keyword list, the competitor set, and the location before work starts, freeze it, and require that any change is flagged in the month it happens. A supplier who resists that is telling you something. Everything else worth reporting follows the same principle, which is the whole argument for an honest monthly report: fixed definitions, stated in advance, by someone who cannot quietly move them later.
Example
Say a mid-sized furniture retailer receives a report showing share of voice climbing for three consecutive months while organic revenue is flat. The set is checked, and two things surface. The tracked list was expanded in month two to include several hundred low-competition terms such as care instructions and assembly guides, where the retailer ranks easily and nobody buys anything. A rival that dominates the high-intent buying queries was also dropped from the comparison after a tool change. Both edits raised the percentage. Neither reflected commercial progress. The fix is not a better metric. It is a frozen keyword set, a stated competitor list, and revenue reported next to visibility every month so the gap between them cannot hide.
FAQ
Is share of voice a Google metric?
No. It is calculated by third-party tools from their own ranking data and their own click-through models. Google publishes impressions and clicks in Search Console for your site only, with no competitor comparison. Any share of voice figure is an external estimate, not something reported by the search engine.
Should brand keywords be included?
Track them separately. Brand terms measure demand you already created, and folding them into the same percentage hides whether non-brand visibility is growing. Two figures, brand and non-brand, answer different questions and are harder to misread. One combined number mostly flatters whoever produced it, and it usually goes unchallenged in the meeting.
How often should the keyword set change?
Rarely, and never quietly. A set should be reviewed perhaps twice a year, when your products or markets genuinely shift. Every change breaks comparability with earlier months, so the report should state what was added, what was removed, and what the previous figure would have been under the new set.
Related terms
- Log File Analysis — a first-party record of crawler behaviour, with no estimation involved.
- Crawl Trap — a structural fault that quietly wastes crawling on URLs no keyword set covers.
- Organic traffic — the counted outcome that share of voice only estimates in advance.
Ask who chose the keyword set before you accept the percentage built on it. If the answer is the same people being measured, you are reading a mark awarded by the candidate.