Brand visibility
DefinitionAnswers that name youAnswers sampled
1,3513,000
=45.0%Aarvi Skin, last 30 daysThe share of sampled answers that mention your brand at all. It is the headline number, and it always shows its count.
Methodology
How Seeqly measures AI visibility.
What we collect, how each number is calculated, and where the limits are. Every figure in Seeqly can be traced back to the saved AI answers behind it.
01 · Collection
How answers are collected.
You choose the buyer questions, or prompts, that matter. Seeqly asks each one every day on every AI assistant in your plan and saves what comes back.
1 · Question“Which sunscreen is best for oily skin?”
A buyer question you track, grouped by topic and intent.
2 · Daily runAsked on each assistant in your plan
ChatGPTGoogle AI OverviewsGeminiClaudePerplexityGrok
3 · SavedFull answer, brands named, sources cited
Kept as evidence, word for word, with the date and assistant.
4 · CountedMetrics built from saved answers
Every number links back to the answers behind it.
Saved answer · ChatGPT12 Sep · United States · English
Question: Which sunscreen is best for oily skin?
“For oily skin, most people pick Dermalume Clear Fluid. Aarvi Skin Matte Gel SPF 50 is a good option if you want a matte finish, and Sunnara is a cheaper choice.”
Brands named, in order
- Dermalume#1
- Aarvi Skin#2
- Sunnara#3
- HydrakindNot named
Sources cited
Review site: best sunscreens for oily skin
Forum thread: matte sunscreen picks
aarviskin.com/matte-gel-spf-50Your page
Example record.
What we keep from each answer
- The full answer, word for word, with the date, assistant, market and language.
- Brands named, in the order they appear, including competitors you track.
- Sources cited, the pages the assistant links to or grounds the answer in. See grounding.
- Claims about you, checked against the facts your team approves, so a hallucination is flagged with the quote.
02 · Metrics
Four numbers, in plain words.
Each metric is a simple fraction of saved answers. The example uses Aarvi Skin, a fictional sunscreen brand, over 30 days.
Share of voice
DefinitionMentions of youMentions of all tracked brands
1,3514,220
=32.0%You plus 3 competitorsHow much of the conversation you hold against the competitors you chose. Adding or removing a competitor changes this number, so we show who is in the set.
Average position
DefinitionSum of your positionsAnswers that name you
4,0531,351
=3.0Lower is betterWhere you appear in the list when you are named. Answers that leave you out are not counted as a position; they show up in visibility instead.
Citation rate
DefinitionAnswers citing your pagesAnswers sampled
1,0003,000
=33.3%Your domain as a sourceHow often an answer uses one of your own pages as a source. Being named and being cited are different, so we report both.
03 · Real counts
Why we show the count, not just a score.
A percentage without its count hides how much evidence is behind it. So every number in Seeqly carries both, like “1,351 of 3,000 answers”, and opens the answers that make it up.
A score on its own72/100
- Which answers? How many?
- Which assistants and dates?
- Can you open the evidence?
A real count45.0%
1,351 of 3,000 answers
- 6 assistants, 1–30 Sep, United States
- 100 questions, run daily
- Click through to every answer behind it
The same kind of number, very different certainty.
A high percentage from a handful of answers says less than a middling one from thousands. Seeing the count lets you judge that at a glance.
Small sample75%
3 of 4 answers
Could plausibly be anywhere from about 30% to 95%.
Large sample45.0%
1,351 of 3,000 answers
Moves only a couple of points either way.
04 · Missing data
“Not measured” is not 0%.
0% means we asked and you were never named. “Not measured” means we have no answers to count yet. Mixing them up makes a gap look like a failure, or hides a real one.
AssistantVisibilityWhat it means
ChatGPT45% 540 of 1,200Named in 540 answers
Perplexity0% 0 of 500Checked 500 times, never named
GrokNot measuredNot in this plan yet, so we have no answers to count
05 · Sampling
Sampling limits and variance.
AI answers are not fixed like a search ranking. Treat every metric as an estimate from a sample, and read trends over weeks, not days.
As a rough guide from basic sampling math, a rate based on 3,000 answers can move about 2 points either way by chance alone; one based on 100 answers can move about 10.
Daily visibility, 14 daysRange 42–49%
The same questions, asked the same way, give slightly different answers from day to day. Sample data.
Answers vary run to run
Assistants do not give the same answer twice. Daily runs build a sample big enough to see the pattern through the noise.
We sample, we do not see everyone
Seeqly asks your tracked questions itself. We cannot see what a specific person was told inside their own account, history or location.
Wording matters
A small change in how a question is phrased can change the answer. Track several phrasings of the questions that matter most.
Compare like with like
Compare periods of the same length, with the same questions, assistants and competitors. Reports flag when the set has changed.
06 · Attribution
What we cannot credit.
A number going up after a change is correlation, not proof. Model updates, competitor moves, seasons and your other campaigns all move AI answers too.
- Impact reports compare the same questions before and after a change.
- They list what else changed in the same period.
- They say plainly what cannot be credited to your change.
- Revenue and pipeline figures depend on connected analytics, store or CRM data. Integrations are rolling out.
Impact report · Matte Gel page update21 Aug–3 Sep vs 5–18 Sep
Answers naming you40% from 20%
Answers citing your page30% from 10%
AI referral revenue+$1,440
What the numbers cannot establish. An ad also changed that week, so the +$1,440 in revenue cannot be credited to the page update. The page change is a likely contributor to the answer shift, not a proven cause.
Recommendation: keep the change, and test again on the next product page. Sample data.
07 · Approvals
Nothing changes without your team.
Seeqly measures and suggests. It does not publish. Every fix is a proposal with its evidence attached, and your team decides what goes live.
- SuggestedA ranked fix, with the answers and sources that justify it
- ReviewedYour team edits the brief or rejects it
- Published by youOn your site, in your CMS, on your schedule
- MeasuredA later sample compared with the one before
Glossary
Terms used on this page.
AI search visibilityAI share of voicePrompt trackingAI citationsAI visibility scoreBrand mentionHallucinationGroundingRetrieval-augmented generation (RAG)AI crawler
GuideWhat is AI search visibility?The idea behind every metric on this page.PlaybookHow to measure AI visibilitySet up questions, competitors and a baseline.FeatureAI share of voiceHow Seeqly reports your share against competitors.FeatureAI citation trackingWhich pages the assistants cite, and for which questions.PlaybookAI search KPIs for CMOsWhich of these numbers belong in a leadership report.PlatformTour the platformWhere each number appears in Seeqly.
How often does Seeqly collect answers?
Every tracked question runs daily on each AI assistant in your plan. Metrics and trends are built from those saved answers.
Which assistants are measured?
ChatGPT, Google AI Overviews, Gemini, Claude, Perplexity and Grok. Your plan decides which ones run; any assistant outside your plan shows as “Not measured”.
Is brand visibility the same as share of voice?
No. Visibility asks “how often are we named at all?” Share of voice asks “of all the brand mentions in these answers, how many are ours?” You can gain visibility and still lose share of voice if competitors gain faster.
Why does my number change when nothing on my site changed?
Assistants update their models and sources, competitors publish new pages, and answers vary from run to run. Look at the count and the trend, not a single day.
Can Seeqly prove a change caused a result?
Not on its own. Impact reports compare before and after, list what else changed in the same period, and say plainly what cannot be credited to your change.
Does Seeqly change my website?
No. Seeqly suggests fixes and drafts briefs. Nothing changes on your site until your team approves it and publishes it.
See your own numbers, with the answers behind them.
Start with Basic free for 7 days, or explore the product first.