Measurement explainer

Answer share (share of answer): what it is and how it is measured

Share of answer is the portion of measured AI answers that include a brand. Collect the answers that AI engines give to a defined set of prompts; share of answer is the number of those answers that name the brand divided by the number of answers collected.

In this page, answer share and share of answer mean the same measure. It is not the same thing as AnswerShare, the company that publishes this page, and it is not a search ranking: it describes how often a brand appears across a defined set of generated answers.

Published 2026-10-18. Last reviewed 2026-10-18.

Talk to AnswerShare Read the case studies

How it is measured

AnswerShare publishes this measure as prompt coverage: how often a brand appears in measured AI answers. The free audit and the methodology use a five-engine panel (Perplexity, OpenAI, Gemini, Claude and Grok). The audit reports a per-engine matrix plus an overall row, and every headline figure is the median across the five models with outlier-drop. A mention is counted separately from a citation, and a crawl is a retrieval event, not an answer.

Report the figure with its prompts, its engines and its date, because engines can diverge and a later answer can differ from an earlier one. The LAVIDGE (now an AnswerShare reseller partner) example below shows the arithmetic on one prompt.

Sources: Share of answer on the methodology page, Free AI visibility audit.

Dated example: LAVIDGE (now an AnswerShare reseller partner), one prompt, five engines

Worked example from a published case study. The signed LAVIDGE (now an AnswerShare reseller partner) receipt records one prompt, “Recommend an ad agency in Phoenix”, put to five AI engines in a fifteen-day audit. The receipt was published on 2026-08-17 14:17:54 UTC and does not record an observation date.

These results were recorded while LAVIDGE was piloting its own site implementation, before it became an AnswerShare reseller partner. They are not results from an unaffiliated client.

LAVIDGE (now an AnswerShare reseller partner), prompt "Recommend an ad agency in Phoenix", signed receipt published 2026-08-17 14:17:54 UTC
Engine	Result	Included in the answer
Perplexity	Ranked #1	Yes
Grok	Ranked #2	Yes
Gemini	Ranked #3	Yes
Claude	Ranked #4	Yes
ChatGPT	Surfaced (included, not ranked)	Yes

Sources: LAVIDGE (now an AnswerShare reseller partner) case study, Signed LAVIDGE (now an AnswerShare reseller partner) receipt, Methodology example.

What the example shows

LAVIDGE (now an AnswerShare reseller partner) appeared in 5 of the 5 engine answers, so share of answer for this one prompt is 5 ÷ 5 = 100%. It was ranked in 4 of the 5 answers (80%). This arithmetic shows how the measure works on one prompt; it is not an AnswerShare score, and a defined prompt set would be larger than one prompt. Engines can diverge, and a later answer can differ from an earlier one.

Sources: Methodology example.

Dated example: V Digital Services (now an AnswerShare reseller partner), with the limits stated

The published V Digital Services (now an AnswerShare reseller partner) record applies the same idea to a different brand. The prompts changed between observations, so the table is a dated record and not a before-and-after comparison.

These results were recorded while V Digital Services was piloting its own site implementation, before it became an AnswerShare reseller partner. They are not results from an unaffiliated client.

V Digital Services (now an AnswerShare reseller partner) dated observations (different prompts; no like-for-like rerun was performed)
Date	Prompt	Engines where the brand appeared
2026-06-17	Enterprise-SEO prompt	0 of 5 (omitted by five of five)
2026-06-17	Phoenix prompt	1 of 5 (omitted by four of five)
2026-07-16	A different Phoenix prompt	3 of 3 measured: Perplexity #1, Grok #1, ChatGPT #2 (Gemini and Claude were not measured)
2026-07-22	Two prompts	ChatGPT did not surface the brand on either

Sources: V Digital Services (now an AnswerShare reseller partner) case study.

Limits of the measure

A crawl is a retrieval event. It is not an answer, a citation or a recommendation.
AI answers are probabilistic, so AnswerShare's repeat-run design asks each prompt of each engine more than once and compares what comes back. A small change in a rate is not evidence of an effect.
One prompt is not a score. A defined prompt set would be larger than one prompt.
Google AI Overviews is tracked by inclusion, cited sources and query coverage, and is not given a chat-engine-style score.

Sources: AnswerShare methodology, AnswerShare FAQ, llms-full.txt.

Published evidence

LAVIDGE (now an AnswerShare reseller partner) case study and signed receipt

The worked example above comes from the published LAVIDGE (now an AnswerShare reseller partner) case study. Signed receipts support integrity: a verifier can tell whether the published payload changed after signing, not whether the underlying observation is true. These results were recorded while LAVIDGE was piloting its own site implementation, before it became an AnswerShare reseller partner. They are not results from an unaffiliated client.

Read the LAVIDGE (now an AnswerShare reseller partner) case study

Receipts and verification

Receipts are published with a standalone verifier and a public key set so anyone can check a payload.

See the receipts

Frequently asked questions

What is share of answer?

What is citation share?

Citation share is the portion of the sources cited in AI answers that belong to a given brand or site. Collect the sources that the answers attribute or link to; citation share is the number that point to the brand's own pages divided by the number of sources cited.

How to measure brand visibility in AI search?

Collect the answers AI engines give to a defined set of prompts, count how many name the brand (share of answer) and how many cite its pages (citation share), and report them by engine with dates. AnswerShare's free audit uses a five-engine panel and reports a per-engine matrix with an overall row.

How to track visibility in AI search?

Repeat the same measurement on a schedule and keep the prompts fixed so the dates are comparable. Because AI answers vary from run to run, track frequency across many runs as a trend instead of a single static rank.

Is share of answer a ranking or a score?

No. It is the portion of collected answers that include a brand. The one-prompt example above shows how the arithmetic works; it is not an AnswerShare score.

Why does the result differ between engines?

Engines behave differently. Gemini names a brand in 83.7% of appearances but generates a clickable citation only 21.4% of the time, and ChatGPT is close to the inverse (Growth Memo/Indig, April 2026). That is why the headline figure is reported per engine before it is combined into a median.

Can I measure my own share of answer?

AnswerShare offers a free AI visibility audit that tests a domain across five engines. Full definitions are on the methodology page.

Talk to AnswerShare

Use the contact page to reach the team. It lists the booking routes for business questions and for technical or press questions.

Go to the contact page