Answers
Answers holds what AI models said about your brand. Mentions lists the answers that name your primary brand, and Responses holds every tracked question with its captured answers. Each answer is one model's response to one prompt at one time, and it is the evidence behind visibility, reputation, content gaps and citations.
Review mentions
Mentions is the default view. It lists primary-brand mentions by capture date, with the surrounding text and the model that produced the answer.
- Choose a date range, segment, or model to narrow the list.
- Open a mention to inspect its evidence.
- Follow the response link to read the full answer in context.
If the list is empty, confirm the brand has a primary brand record, widen the date range, and check that prompts have captured answers. An answer that does not mention your brand is still available under Prompts.
Read a prompt
Open Answers → Responses and choose a prompt. Its page shows how each engine answered it over the period in the page bar. The period is the last 30 days unless you choose another, and the answers you open from the page keep it.
The first line sums up the period. It says which tracked brand wins the prompt, how many engines named and recommended you in their latest answer, and how often you were named overall. It says a figure rose or dropped only when it changed by more than chance since the same number of days before.
Three numbers follow. All adds up every engine the segment asks, and the tab for an engine shows just its answers:
- Named you counts the answers that mention your primary brand.
- Recommended you counts the answers that recommend it. An answer still being analysed is left out until its analysis finishes.
- Cited your site counts the answers that cite a page on your website.
Below 10 answers, the numbers show counts such as "3 of 4" instead of a percentage, because a percentage of a few answers can swing a lot from one run to the next. When Google showed no AI answer for the question, that run counts as an answer that named nobody, and Google's tab says how many runs that happened in. An engine the segment no longer asks has a greyed-out tab marked No longer asked. Its answers count only in its own tab, not in All.
Who wins ranks your brand and its competitors by how many answers name them, then by their average position in those answers. A lower position means the engine named the brand earlier.
Run by run shows each engine's answer in every run, oldest on the left. Scroll sideways to see older runs. When two runs fall on the same day, their headings also show the time. Each cell shows what that answer did:
- A number is your brand's position in the answer. ✓ means it named you without a position, and × means it didn't name you.
- A filled cell with ★ means the answer recommended you.
- A dashed cell with … means the answer is still being analysed, so it isn't known yet whether it recommended you.
- A dot means the answer cited your site.
- ○ marks a run where Google showed no AI answer, and ! marks a failed run.
An engine the segment no longer asks is greyed out and marked No longer asked. Select a cell to read that answer.
A prompt in a reputation segment feeds Reputation and doesn't count towards visibility. Its page shows the engines and the runs, without the first line or Who wins.
Read an answer
Open Answers → Responses, choose a prompt, and select an answer in its Run by run grid. You can also follow a link to an answer from Mentions, Opportunities or a Reputation narrative. Each answer has its own page, so you can bookmark it or send its address to someone with access to the brand. Older links to answers still open the right answer.
The tabs above the answer show each engine's answer to the same prompt in the same run. Select a tab, or press ← and →, to switch engines. A tab with no answer says why:
- No AI answer means Google showed no AI answer for the question.
- Failed means the engine's run failed.
- Not in this run means the run didn't ask that engine, or its answer hasn't arrived yet.
Switching engines doesn't add to your browser history, so Back returns to the prompt in one step.
The run menu in the page bar lists the prompt's runs, newest first. It covers the period of the page you came from, which is the last 30 days unless you chose another. For an older answer it covers the same number of days up to that answer. Choose a run to read the same engine's answer from it.
Run details lists every setting and measurement Cleotic recorded for the run, such as the model version, the tokens used and how long it took. Copy as JSON copies them all, which helps when you contact support. Answers collected before Cleotic recorded runs have no run details.
Check the model and capture time before comparing answers. Repeated runs of the same prompt can produce different wording or recommendations. To see how each run asks the engine, see How answers are collected.
A line above the answer says what the engine did to write it. See What the engine did.
Use Text for formatted text or Raw to inspect the captured wording. Detected brand mentions can be highlighted or opened for more detail. Where an engine points a sentence at a source, Text shows a small pill after it with the source's icon and site name; select it to open the source. Several sources from the same site after one sentence share a single pill.
Below the answer, Citations lists the sources the engine reported, with their details, and Links in the answer lists the other links the answer contains. See Citations and links in the answer.
Notes beside the answer include, where available:
- Claims about your brand, on Business and above. See Claims about your brand.
- Response classifications and the reasons for them.
- Brand mentions and their positive, negative, or neutral sentiment. A mention also shows the site the answer linked the brand's name to, when it did.
- Recommendation status and ranked recommendations.
Analysis can be pending or unavailable even when the answer text is ready. Read the captured answer directly when a classification is missing. A missing classification is not a negative result.
Mentioned is different from recommended
A brand can appear in an answer as an example, an alternative, or a criticism. Use the surrounding text and recommendation evidence to understand what the model actually suggested.
Claims about your brand
On Business and above, an answer that names your primary brand shows Claims about your brand beside it. Each statement the answer makes about your brand is marked in the text with a number. Its note says what Reputation found, such as Matches your facts, Contradicts your facts, Out of date or Contact detail not on file. The mark is red for a risk, amber for a warning, green for a match or a positive claim, and a grey dotted line when there is no verdict.
Select a number to jump to its note. A note shows the confirmed fact and the reason where there is one. Open narrative opens the narrative the claim belongs to, and Add a fact opens Brand settings → Facts for a claim or contact detail that no fact covers. Checking claims means the check hasn't finished yet.
What the engine did
The line above the answer shows what the engine reported doing while it wrote the answer, such as "Searched 2 times · Read 7 pages · Cited 5 sources". An engine that doesn't report a step leaves it out of the line, so a missing step doesn't mean zero.
- Searched says how many web searches the engine ran. Select it to see what it searched for, in order. Engines often rephrase your question into their own searches, so this shows the wording a page needs to match to be found.
- Gemini reports how many searches it ran, but not what it searched for.
- Google AI Overviews and Google AI Mode don't report their searches, so the line doesn't mention searching.
- "Answered without searching" means the engine answered from what it already knew.
- Read appears only for Claude, the one engine that lists the pages its searches returned. When it read pages it didn't use in the answer, select Read to see them under Read but not cited. A page of yours in this list was found but not chosen, which often means it didn't answer the question directly enough.
- Cited counts the sources the engine reported. They are listed under Citations below the answer.
Citations and links in the answer
An answer can contain two kinds of link, and Cleotic keeps them apart.
- Citations are pages the engine reported as sources for the answer. These feed AI citations and every citation count.
- Links in the answer are links the engine wrote into the text without reporting the page as a source. Google often links a company's name to its website, and models sometimes point you to a page to visit.
A link in the answer means the engine named that company or page, but not that it relied on the page. Citations don't reveal every source in the model's training or prove that the source supports the claim. Open the source when the claim matters. See AI citations.
History and changes
Replacing a prompt's wording keeps earlier answers under the old prompt. Editing brand names or aliases can change historical matches after reanalysis, while the captured answer stays the same. See Prompts.
An expired trial stops new capture and makes the workspace read-only. Access to older history also follows your plan's retention window. See Plans and usage.

