Share of Voice (SOV)
Your slice of all brand mentions in your prompt set.
Last updated: 2026-05-05
Share of Voice (SOV)
Share of Voice answers a different question than AI Score: of all the brand mentions in AI answers about your category, what percentage belong to you? It's a relative, competitive metric — your SOV can grow even if your absolute mention count shrinks (because competitors lost more), or shrink even if you grew (because a new entrant grew faster).
How it's calculated
For the active filters (date range, country, prompt type, models):
- Pranas counts every brand mention across every query in scope.
- URL slugs are deduplicated — if a single AI answer mentions "pranas.co" three times, it counts as one mention. This prevents verbose models from inflating SOV.
- Child brands roll up to parents — if Brand A has child Brand A-Pro, mentions of A-Pro contribute to A's SOV.
- The result is
your_mentions / total_mentions_in_category × 100.
The "category" universe is defined by your registered competitors plus any unregistered brands the AI surfaces in your prompts. Unregistered brands are bucketed as Other by default; toggle Competitors Only to exclude them.
Reading the SOV chart
The dashboard shows SOV as a stacked time-series:
- Your brand is always on top in your brand colour
- Registered competitors stack below in their assigned colours
- Unregistered/Other sits at the bottom in muted grey
A rising your-share segment with a shrinking Other segment is the cleanest growth signal — you're capturing demand from previously unranked players, not stealing from a known competitor.
SOV by prompt type
| Prompt type | What that SOV tells you | Primary audience |
|---|---|---|
| Unaided | The closest thing to true market share in AI answers | Executives and investors |
| Comparison | Head-to-head win rate; drops here often precede pipeline drops | Sales and product marketing |
| Problem-Solving | How well your content answers real jobs to be done | Content marketing |
| Aided | Recall when the prompt already names you; inflates your own share | Diagnostics only |
SOV versus Top Brands
| Aspect | Share of Voice | Top Brands |
|---|---|---|
| Counts | Total mentions | Distinct queries naming the brand |
| Rewards | Prominence within answers | Reach across answers |
| Deduplication | URL slugs deduped per answer | One count per query |
| Best for | Competitive share reporting | Coverage and rank reporting |
Segmenting SOV by prompt type
The most actionable view is per prompt type:
- Unaided SOV — the "true" market share signal. This is what investors and execs care about.
- Comparison SOV — head-to-head win rate. Drops here predict pipeline drops.
- Problem-Solving SOV — content marketing scorecard.
Use the prompt-type filter at the top of the dashboard to switch.
Why SOV may differ from Top Brands
Top Brands counts how many distinct queries name each brand. SOV counts total mentions. A brand mentioned twice in every query will rank higher in SOV than in Top Brands. Both views are valid; SOV rewards prominence, Top Brands rewards reach.
Common pitfalls
- Tiny prompt libraries (under 10 prompts) produce noisy SOV. Add prompts before drawing conclusions.
- Mixed countries can skew SOV because some competitors are regional. Filter by country for cleaner comparison.
- Including Aided prompts inflates your own SOV because the prompt names you. Default to Unaided for fair benchmarking.
Related articles
- How Pranas measures AI visibility
- Top Brands widget
- Co-occurrence analysis
- Country scoping