> ## Documentation Index
> Fetch the complete documentation index at: https://docs.peakmark.cc/llms.txt
> Use this file to discover all available pages before exploring further.
# AI Share of Voice: See Who AI Recommends in Your Category
> AI share of voice shows what percentage of relevant AI answers mention your brand versus competitors. Track it over time to measure your GEO progress.
Knowing whether your brand appears in AI answers is a start. Knowing how that appearance compares to every competitor in your category — across hundreds of prompts and four major AI models — is where strategy becomes possible. AI share of voice (SOV) is that competitive picture. It moves the conversation from "am I visible?" to "how visible am I relative to the brands I'm actually competing against, and where is the gap I can close fastest?" For teams investing in GEO, share of voice is the primary metric that proves whether the work is winning.
## What Is AI Share of Voice?
**AI share of voice** is the percentage of category-relevant AI answers that name your brand, measured against the same percentage for your competitors.
More precisely: across all the prompts Peakmark runs in your category, how many responses include your brand versus how many include Competitor A or Competitor B? If 100 prompts are run and your brand appears in 36 of them, your share of voice is 36%. If your closest competitor appears in 54 of the same 100 prompts, their SOV is 54% — and you have a 18-point gap to close.
## Why Share of Voice Is the Right Metric
A search rank is a snapshot of a single keyword at a single moment. It tells you very little about competitive dynamics and nothing about how buyers across an entire category of questions experience your brand versus alternatives.
AI share of voice is different in three critical ways:
* **It's competitive by definition.** SOV is always relative — your percentage versus theirs, not an absolute score that can be manipulated in isolation.
* **It aggregates across many prompts.** A single AI response is noisy. SOV calculated across hundreds of prompts gives you a statistically meaningful signal.
* **It tracks over time.** A line chart of your SOV over 90 days tells a story that no weekly rank report can: are you gaining on the leader, holding steady, or losing ground?
## How Peakmark Calculates Share of Voice
Peakmark's SOV calculation is transparent and reproducible:
1. **Prompt set definition** — Peakmark builds a curated set of prompts that represent the questions your buyers actually ask about your category. These prompts are reviewed and refined to ensure they reflect real buyer intent, not synthetic queries.
2. **Daily model runs** — Every prompt in the set is sent to each tracked AI model every day. Responses are captured in full.
3. **Brand identification** — Each response is analyzed to identify every brand mentioned, regardless of position. Both primary recommendations and secondary mentions are counted.
4. **SOV calculation** — For each brand:
```
Share of Voice = (prompts mentioning your brand ÷ total prompts) × 100
```
This is calculated per model and in aggregate across all models, giving you both a headline number and the model-by-model breakdown.
## What the Dashboard Shows
The Peakmark SOV dashboard surfaces competitive intelligence at four levels:
* **Your SOV over time** — a line chart showing your share of voice by day or week, with the ability to overlay competitor trends on the same chart
* **Top competitors and their SOV** — ranked automatically by share of voice so the dominant player in your category is always visible
* **Model-by-model breakdown** — your SOV on ChatGPT, Claude, Gemini, and Perplexity tracked separately, so platform-specific gaps are visible
* **Prompt-level evidence** — click any data point to read the actual AI responses behind the number, so you know exactly how your brand and competitors are being described
## Example: SOV Breakdown by Model
The following illustrates a typical competitive snapshot for a mid-market SaaS brand. Note how model-level performance varies significantly even for the same brand.
| Brand | ChatGPT | Claude | Gemini | Perplexity | Avg |
| ------------ | ------- | ------ | ------ | ---------- | --- |
| Your brand | 42% | 38% | 29% | 35% | 36% |
| Competitor A | 58% | 61% | 45% | 52% | 54% |
| Competitor B | 21% | 18% | 37% | 24% | 25% |
In this example, your brand trails Competitor A by 18 points on average. But the gap is largest on Claude (23 points) and smallest on ChatGPT (16 points). Competitor B leads you on Gemini by 8 points despite trailing overall. Each of these observations points to a different prioritization decision.
## Using SOV to Prioritize Your GEO Work
Share of voice data answers the most important resource allocation question in GEO: *where should I focus next?*
The framework Peakmark recommends:
Identify the model where the difference between your SOV and the category leader's SOV is greatest. That's where the most incremental value is available.
A large gap on a model that indexes heavily on earned media may be more actionable in the near term than a gap on a model that weights long-term training data. Use the citation source report to understand what's driving the leader's advantage.
Pick one model, set a SOV target — e.g., close the gap by 10 points over 90 days — and align your content, PR, and community efforts to that specific model's citation patterns.
SOV moves on a timeline of weeks, not days. Check the trend weekly. A consistent upward slope is evidence your work is landing; a flat line after 30 days is a signal to revisit the tactic mix.
Don't optimize for aggregate SOV alone. Two brands can have identical average SOV while having very different model-level profiles. Always check the per-model breakdown before deciding where to invest.
## From Metric to Action
Share of voice is a measurement, not a strategy. Once you know your gaps, the next step is understanding which competitors are winning citations on specific models and what sources are feeding those citations. Peakmark's competitor analysis view links SOV data directly to the source citations behind each brand's mentions — so you can identify the earned media, Reddit threads, and review profiles that are actually driving the gap, and build a plan to win them.
See exactly which competitors AI recommends in your category, what sources drive their citations, and how to close the gap — prompt by prompt, model by model.