Now the model chooses
It reads, summarizes and decides which brands fit the answer. There's no auction, no bidding, no way to buy the spot: whoever it can justify with what it found gets in.
All services
The eyes · AEO/GEO
Someone asked AI about your industry today. You weren't in the answer.
“It's not a suspicion: it can be measured. The module asks your market's questions in the assistants where your customer asks, saves what came back, and shows which questions your brand appears in, which ones it disappears from, and who is taking your place.
MEASURED TODAY INChatGPTGeminiPerplexityClaude
A CUSTOMER, RIGHT NOW
What's the best [your industry] company for an operation my size?
ChatGPT
ChatGPTGeminiPerplexityClaudechanging engines changes the answer, and that's why measurement happens on each one
Nobody warns you when you're left out.
With the old search engine you could see the decline: your ranking fell, your traffic fell, someone noticed. In the AI's answer there's no second page and no ranking drop. There's only being in or being out. And the customer who never saw you also never knew you existed to search for.
What changed
The question no longer returns ten links. It returns three names.
You're not penalized. You're omitted
There's no alert, email or dashboard telling you your brand dropped out. The answer simply moves on without you, and the conversation ends before it reaches your site.
Each engine has its own version
The same question brings up different brands in ChatGPT, Gemini, Perplexity and Claude. Doing well in one says nothing about the others, which is why measurement needs to happen engine by engine.
What you start to see
Six numbers nobody at your company can answer today.
Each round the module asks, reads the whole answer and stores what it found. What you open in the morning isn't an impression: it's a picture of what the AI has been saying about your brand, with a date and the source next to it.
EXAMPLE OF A SNAPSHOT
0%
Visibility0%
Reputation and tone0º
Average position0
OpportunitiesVisibility
Out of every ten questions in your market, how many mention your brand. It's the number that sums everything up, and the only one you can track month by month.
Below a competitor, you already know where to attack
Reputation and tone
Being mentioned isn't enough: how matters. The module separates praise from caveats and from comparisons where you show up as the pricier option.
A caveat that keeps repeating is content that's missing
Position and share of voice
Where in the answer you show up, and how much of the space among cited brands is yours compared to each competitor's.
Shows who's winning the category
Sources and opportunities
Where the AI got what it said, and the list of questions you didn't appear in, which is where the next work is.
Each gap becomes a task with an owner
Engine by engine
Four assistants, four different answers.
Measurement is done on each one separately, with the question written the way a customer would write it. No average that hides the engine where you're doing poorly.
ChatGPTmeasuring
The one with the largest reach, and the one that mixes memory with search the most. It answers by citing few brands and tends to repeat the same ones: whoever gets in, gets in to stay.
Geminimeasuring
It pulls heavily from what Google already indexes, so it rewards well-structured pages and pages that answer the whole question, not just the title.
Perplexitymeasuring
It always cites the source right in front of the user. It's the engine where you can most clearly see WHICH page made the AI talk about you, or about your competitor.
Claudemeasuring
More conservative: it prefers to state what it can back up. Appearing here usually requires third party material, not just your own site.
Google AI Modecoming soon
The answer Google builds right into search. The collector already exists and is being calibrated. It goes live for customers as soon as the reading is verified.
An engine only starts counting once the collection is checked. Doubtful data doesn't become a metric here: we prefer the smaller, true list.
No sign-up
Find out right now if your brand shows up.
A sample measurement, run on the spot, on the engines that are up. You type in your site and see the result on this page: signing up only makes sense after you know the number.
To take with you
The ten questions that decide your sale.
Type in your industry and take the ready-made list. Paste it into ChatGPT tonight and see with your own eyes who gets cited: that is exactly what the module does, every day, on four engines.
- 01
What's the best [your industry] company?
- 02
What are the main [your industry] options in Brazil?
- 03
Is it worth hiring [your industry]? What are the pros and cons?
- 04
How much does [your industry] cost and what's included?
- 05
What's the cheapest [your industry] alternative without losing quality?
- 06
Compare the three best known [your industry] companies.
- 07
Which [your industry] company is best for a mid sized operation?
- 08
What should I watch out for when hiring [your industry]?
- 09
Who are the most trustworthy [your industry] companies?
- 10
What's the best [your industry] option for someone just starting out?
These are templates, not measurement: they serve as a starting point. In the product, questions come from your market and what your customers actually write.
To settle the question
The four comparisons nobody makes out loud.
None of them caricature the other side: SEO works, spreadsheets work, and chart tools are good for something. The point is where each one stops.
SEOAEO/GEO
What's measured Position in a list of links Whether the brand is mentioned, and how
Who decides The search engine index The model, before any click
What makes it change Title, link, page speed Material that backs up a claim
How you find out it dropped Traffic drops and someone notices Nobody notices, unless you measure
One question is the difference between existing and not existing
Without media budget, organic citation in the answer is the channel. Start by measuring five questions and writing for the two where you don't appear.
There's already someone producing content: what's missing is where to point it
The team publishes, but nobody knows if that turned into a citation. Here the material comes from the measured gap, and the next round tells you if it worked.
Multiple brands, multiple markets, the same dashboard
Track line by line, by product and by region, with exportable history and the memory staying with the company.
Inside
Why AI cites one brand and not the other.
It's not a raffle or an auction. The model needs to back up what it says, and it picks what it can justify with what it found. Four things matter more than the rest.
Someone outside talked about you
A mention in a media outlet, comparison, or forum is worth more than your own site: the model treats a third party as a witness and you as an interested party.
what to do: appear in industry lists and comparisons
Different sources say the same thing
When three places describe you the same way, the model gains confidence to state it. A description that changes with each channel weakens the citation.
what to do: one single description, repeated everywhere
The page answers the whole question
Text that answers directly, with the question in the title and the answer right below, is easier to cite than a nice looking institutional page.
what to do: one page per real customer question
It's recent enough
In a category that changes fast, the model favors this year's material. A 2019 page with no update loses to a smaller competitor's.
what to do: review the date and figures of what already exists
From measurement to work done
A report doesn't change the AI's answer. Published work does.
This is where most visibility tools stop: they deliver the chart and leave. Here, the gap found already comes out as a task, and the task already comes out as published material, because published material is what the next round will find.
Measure my brand for freeVISIBILITY · SIX ROUNDS (EXAMPLE)18% → 63%
Each step is a piece of content published from a measured gap. The chart doesn't rise on its own: it rises when someone writes what was missing.
01
The question that left you out
The module shows the exact question, the engine it was asked on, who was cited in your place, and the page that backed up that citation.
02
Becomes a task in the action plan
With what needs to exist, where it needs to exist, and the priority: the gap that appears in three engines is worth more than the one that appears in one.
03
And the next round checks
Two weeks later the same question is asked again. Either your brand made it into the answer, or it didn't, and that shows up in the chart, not in a meeting.
From zero to the first snapshot
Three steps. The first number on the same day.
012 MINUTES
Register the brand and the sector
Name, website, and which market you're in. That's how the module knows which questions make sense.
0210 MINUTES
Choose the questions
The system suggests the ones your customer would ask; you remove, add, and select two to four competitors to compare.
03SAME DAY
See the first snapshot
The first round comes out right away. From then on, it repeats on its own, at the frequency you chose.
Inside the company
The same measurement, three readings.
Stop writing in the dark: the topic comes from the question where the brand didn't show up, and the result of the piece appears in the next round.
Find out which competitor the AI is delivering to the customer before the first meeting, and with what argument.
See how the market describes the category and where your proposition is being misread, without waiting for quarterly research.
What comes out of this, concretely
The dated snapshot
Visibility, reputation, position, sources, and opportunities: exportable.
Page audit
Why that page doesn't become a source, and what's missing from it.
Article written to be cited
The material that fills the gap, built from what the engines showed.
Where the alert arrives
No one needs to remember to open another dashboard. When the brand drops out of an answer, the alert reaches where the team already is.
Slack Email WhatsApp
View Connect How much it costs
You pay per measurement, and the engine sets the price.
Web search costs different prices on each engine, and Rovemark passes on the difference instead of charging the average. Whoever wants volume measures on Perplexity; whoever wants Gemini's answer pays what it costs. The choice is up to whoever pays.
One question on one enginePerplexity 20, ChatGPT 30, Gemini 100, Claude 130. The difference is each one's search price, not our margin.
20 credits
Audit of one pageWhy that page isn't being cited and what it's missing to become a source.
~80 credits
Article written to be citedThe material that fills the gap, written from what the engines showed is missing.
~120 credits
the questions your customer would ask before buying
Engines measuredeach engine charges what its search costs: 20 on Perplexity, 130 on Claude
How oftendaily shows the change in the week it happens
SPEND PER MONTH45,000credits
900 measurements, each with the full answer stored
The free plan starts with 1,000 credits a month. On Perplexity that's 50 measurements; on Gemini, 10. You can find out if you're in or out without a card, choosing where to spend.
Measure my brand for freeBefore you ask
What people usually want to know.
Is this SEO with another name?
No. SEO competes for position in a list of links the user still has to choose from; here the choice has already been made by the model before any click. It changes what is measured (citation, not position), changes who decides (the model, not the index) and changes the work: what gets AI to cite you tends to be material that backs up a claim, not an optimized title. Whoever already does SEO well starts ahead, but it's not the same thing.
How often does the measurement run?
You choose: hourly, every six hours, twice a day, daily or weekly. The default is daily, which is enough to see change without wasting credit: an AI's answer doesn't change from morning to afternoon, it changes when someone publishes something.
How many brands can you track?
As many as your balance can handle: the count is per question measured, not per registered brand. In practice most people track their own brand and two to four competitors, because comparison is exactly what gives the number meaning.
I already have an agency. Does this get in the way?
It helps. The agency starts working with the exact question that left you out and the page that got your competitor cited, instead of trying to guess. And the result of what it publishes shows up in the next round, dated.
What about my data?
What your company accumulates here is stored in Cerne with your key: questions, collected answers, history. It comes out whole the day you want it, and it never becomes anyone's model training data.
One line each
The vocabulary, straight to the point.
Definitions short enough to be cited, including by an AI asked about it.
- AEO
- Answer Engine Optimization: the work of getting your brand cited in the answer an AI assistant gives, instead of competing for position in a list of links.
- GEO
- Generative Engine Optimization: the same goal, stated from the point of view of the generative engine that produces the answer. In practice, AEO and GEO are used as synonyms.
- Share of voice
- How much of the space of brands cited in an answer is yours, compared to each competitor's. Twenty percent means one in every five mentions is your brand.
- Citation
- The mention of your brand's name inside the answer. It's the unit AEO measures, with no citation there's no click, because the conversation ends at the answer.
- Source
- The page the model used to back up what it stated. Finding a competitor's source means finding exactly the material missing on your side.
- Gap
- A question from your market where your brand isn't cited. It's where the next sale is, and it's what the module turns into a task.
THE OTHER ORGANS
No module works alone.
Each one solves a front, and all write to the same memory. See the others and how they fit with this one.