Early access
Start free trial
Glossary · Metric

What is share of voice (AI)?

The classical marketing benchmark adapted for AI answers — how a brand stacks up against a named competitor set inside the responses engines generate.

Last updated June 2026

In detail

Share of voice (AI) adapts the classical share-of-voice metric — historically a brand's slice of category advertising — to AI search answers. The computation requires two stable inputs: a defined competitor set, typically three to seven brands, and a fixed prompt list, so the numbers stay comparable over time.

It pairs with share of model on most dashboards. Share of model is the primitive — absolute visibility per engine. Share of voice is the relative benchmark — visibility against the competitive slate. Reading them together prevents the classic mistake: celebrating rising share of model while losing ground to competitors growing faster.

For ecommerce specifically, the brand-level version has a sharper SKU-level twin: define the set as competing products rather than competing logos, count product citations rather than name-drops, and the metric starts answering merchandising questions instead of brand-marketing ones.

Why it matters for ecommerce

A brand cannot control the size of the AI answer window. ChatGPT might cite three brands this quarter and five the next for the same prompt. Share of voice normalizes for that: the denominator is mentions within the competitor set, so the number stays comparable even as engines change their answer style.

For a VP of SEO, it is also the most familiar AI metric in the stack — the closest analog to the rank-tracking and visibility-share reports already in the monthly deck. It turns "are we winning in AI answers" into a number leadership recognizes, with the competitive frame built in.

Example

A merino base-layer brand defines a five-brand competitor set and tracks a 30-prompt bank across ChatGPT and Google AI Overviews. In Q1 the brand earns 9 of 45 total set mentions — 20% share of voice. Two competitors pick up review-site coverage in Q2, and the brand slides to 6 of 52 mentions — 11.5% — even though its absolute mention count barely moved. The share-of-voice drop flags the competitive shift two months before it would have surfaced as a revenue dip.

How eCommerce Insights computes it

Share of voice is computed from the same retained prompt-tracking runs as share of model — per engine, per week, against a competitor set the team defines once and can hold stable. Because tracking resolves to SKUs, the product-versus-product version comes free: the dashboard shows which competitor PDP took the citation, not just which logo. See SKU-level tracking.

Related terms

Ask AI about share of voice (AI)

Have your favorite AI engine apply this definition to your catalog.

Frequently asked questions

How is share of voice (AI) different from share of model?
Share of model is absolute — how often one brand appears out of total responses on one engine. Share of voice (AI) is relative — how often one brand appears out of all mentions across a named competitor set. Share of model says the brand is in 18% of ChatGPT responses; share of voice says the brand holds 18% of the mentions that went to the five-brand competitive set.
How do I choose the competitor set?
Pick the brands a shopper actually weighs in the category — typically three to seven: direct competitors plus the one or two category leaders that keep showing up in answers regardless. Keep the set stable for at least a quarter so the metric stays comparable; adding or removing a competitor resets the baseline.
Do I need both share of model and share of voice?
Yes, because each catches what the other misses. Share of model can rise while competitors rise faster — share of voice exposes that. Share of voice can hold steady while the whole category loses AI coverage — share of model exposes that. Reading them together is the standard dashboard pairing.
Why does share of voice stay comparable when engines change their answer style?
Because the denominator is mentions within the defined competitor set, not the size of the answer. ChatGPT might cite three brands this quarter and five next quarter for the same prompt; the share-of-voice math divides by total set mentions either way, so the number remains comparable across weeks even as answer formats shift.
Can share of voice be computed at the SKU level?
Yes, and for ecommerce it should be: define the competitor set as rival products rather than rival logos — the five stability running shoes that contest the same prompts — and count product citations instead of brand mentions. That is the version that tells a merchandiser which specific competitor product is eating the answer slot.

Go deeper

Benchmark your share of voice against the competitors that matter, per SKU. Start the 14-day free trial.