On Google, the advantage goes to those who spent a decade building domain authority and links.
at its peak2.3bn
now1.7bn
monthly visits that organic search still sends to content sites. That's 600 million fewer
similarweb · 2026
free diagnostic · 95 seconds
How your brand shows up
How your brand shows up
when people ask AI
We read your site, write the questions people ask in your market, and run them across 4 engines.
chatgpt
claude
perplexity
ai overview
You don't have to keep watching.
I don't want to wait Leave your email: the analysis continues while you're away. the run stopped
You have not been charged, and no data has been lost. The work already completed is recorded below. Try again in a few minutes.
if it happens again, send the code above to
account@murmur.marketing
if it happens again, write to
account@murmur.marketing
We put your brand in AI answers.
3%31%
murmur is an AI visibility engine
We measure where your brand appears in model responses and where it doesn't, then produce the content that changes the picture. Data decides what gets written.
One person dedicated to your brand
You get a specialist who knows your industry, follows your results and takes responsibility for them. The same person, from the diagnostic onward.
A study built around your business
Every question in the study comes from your market and is chosen individually, from early research to the decision to choose a provider.
We handle production
We write and publish on a subdomain of your brand that we build and maintain. Your main website stays untouched: no access to your systems and no risk of breaking pages that work.
The plan follows the measurement
There is no fixed package. The number of questions, the amount of content and the publishing pace all follow from the diagnostic, once the actual scope of the work is clear.
In AI answers, the advantage goes to those the model can understand, trust and cite.
Feb 2025400m
Feb 2026900m
people using ChatGPT every week, twice as many in one year
openai · Feb 2026
the engine
Four stages, one continuous cycle.
With each cycle, the strategy adapts and more questions in your market are answered with your brand's name.
recommended
cited
absent
-
01 · mapping
We map the questions people ask in your industry, from early research to choosing a provider. We ask them in the actual product, capturing and storing each response.
-
02 · diagnosis
We read every answer and record where your brand appears, where it doesn't, and which competitor takes its place when it is absent.
-
03 · content and structure
Every piece starts with a measured gap and takes a form the models can cite. We publish it with markup that tells the model what each passage is.
-
04 · continuous monitoring
Measurement continues after publication and determines the next step. An upcoming seasonal campaign, a competitor gaining ground on a question, or a topic that hasn't gained traction changes the following week's priorities.
- and the cycle starts again with an updated map
Tracking dashboard
Your share of voice alongside each competitor's, measurement by measurement. Everyone gets the same questions on the same day. That is what makes the comparison fair.
How often AI recommends you, cites your domain or lists you first, and how each figure breaks down across the four engines.
The full list of active prompts, with visibility for each one, the number of competitors in that answer and its stage in the funnel.
Your share of voice alongside each competitor's, measurement by measurement. Everyone gets the same questions on the same day. That is what makes the comparison fair.
murmur · geo trackingtimelineyour-brand
share of voice
- competitor A37.6%
- your brand36.8%
- competitor B29.4%
- competitor C27.0%
- competitor D22.6%
- competitor E11.8%
visibility by brand
captured answers · over time
best companies in the industry
Jun 27
which provider to choose
Jun 29
who leads this market
Jul 22
is it worth hiring
Jul 24
comparing the leading providers
Jul 31
which company serves businesses my size
Aug 12
who would you recommend
Aug 19
How often AI recommends you, cites your domain or lists you first, and how each figure breaks down across the four engines.
murmur · geo trackingmetricsyour-brand
perception scale
absentcitedrecommendedbenchmark
20.0%
recommendation · 5 points to benchmark
AI cites your domain40%
non-negative sentiment99%
listed first58%
unprompted brand recall6.1%
by engine
- chatgpt19.1%
- claude5.6%
- perplexity37.6%
- ai overview0.0%
The full list of active prompts, with visibility for each one, the number of competitors in that answer and its stage in the funnel.
murmur · geo trackingpromptsyour-brand
313 of 313 prompts · latest complete run
| prompt | competitors | stage |
|---|---|---|
| best companies in the industry in Brazil | 3 competitors | bottom of funnel |
| which provider to choose when starting out | 1 competitor | top of funnel |
| who leads this market today | 4 competitors | middle of funnel |
| is your brand worth hiring? | 1 competitor | bottom of funnel |
| comparing the leading providers | 5 competitors | bottom of funnel |
| which company best serves businesses my size | 2 competitors | middle of funnel |
Cases
We measure, do the work, then measure again using the same yardstick. Each case starts with the full figure, its denominator and what didn't work.
Mindo · B2B motion design studio in São Paulo#1when ChatGPT lists the best animation studiosIn May 2026, it wasn't on the list.
Read the case ›FotoRegistro · online photo printing and personalized photo products#1when ChatGPT recommends where to print photos onlineIt used to appear, but as the cheap option.
Read the case ›Fly Vet · veterinary marketing agency with its own technology+78%Average deal value. SQL conversion at 68%.It previously depended on paid traffic to make sales.
Read the case ›Visio · operations software for multi-store retailers, franchises and quick-service restaurant chains193answers about retail management where it already appearsIn June 2026, AI didn't even know it existed.
Read the case ›
Read the case ›FotoRegistro · online photo printing and personalized photo products#1when ChatGPT recommends where to print photos onlineIt used to appear, but as the cheap option.Read the case ›Fly Vet · veterinary marketing agency with its own technology+78%Average deal value. SQL conversion at 68%.It previously depended on paid traffic to make sales.Read the case ›Visio · operations software for multi-store retailers, franchises and quick-service restaurant chains193answers about retail management where it already appearsIn June 2026, AI didn't even know it existed.Read the case ›The technical foundation
semantics · how the text is written
a verifiable source for every claim
A claim without a source is discarded. A claim with a source the model can verify is what it repeats when it needs to support an answer.
quotes from authoritative sources
A passage attributed to a named person or institution is the format most likely to increase the chance of the whole passage being reused.
data instead of adjectives
"Market leader" tells a machine nothing. A number with a unit and a time period does.
the language people use to ask
The text answers in the words the person used, rather than the company's internal jargon. "Integrated management solution" loses to "inventory management software."
clarity first
Short sentences, explicit subjects, one idea at a time. What is ambiguous to a reader is impossible for a model to extract.
and what stopped working
Repeating keywords, a tactic that sustained twenty years of SEO, does not move AI answers. Sometimes it gets in the way.
data structure · what goes on the site
explicit schema markup
Invisible labels that tell a machine what each part of a page is: an article, a frequently asked question, an author.
llms.txt
A site index written for language models: what is there, what matters most and where to find it.
explicit access for crawlers
The crawlers that read the web for AI, including ChatGPT, Claude and Perplexity, are explicitly allowed by name. Most sites block them without knowing it.
content served as markdown
When a bot requests a page, it receives just the text, without menus, buttons or images. It arrives cleaner and faster.
conversational search
A field where people can ask in their own words and receive an answer instead of a list of links.
an introduction for agents
A file in a standard location, where agents know to look, explaining what the site offers and how to request it.
Questions
It depends on how well the model already knows your industry and how many questions are being contested. The initial measurement shows your position in the first week. The first citations tend to appear in the weeks after publication, rather than months. Scale takes longer: covering hundreds of questions is ongoing work.
We do. The content goes on a subdomain of your brand, such as content.yourbrand.com, which we build and maintain. Your main site stays untouched: no access to your systems and no risk of breaking pages that already work. What we build stays on your domain, not ours.
We run the same set of questions across the four engines, week after week, and keep a captured copy of each answer. These are the AI's actual responses, recorded and dated. When we say your brand is now cited for a particular question, you can see the answer.
The content and structure we produce apply to any model that reads the open web, because optimization is not exclusive to one model. Measurement coverage is a separate matter: we track ChatGPT, Claude, Perplexity and Google's AI Overview, where most questions are concentrated.
There is no fixed price list. Pricing is built around each brand's plan: how many questions the study covers, how much content is needed and how often it should be published. We make the proposal after the diagnostic, when the actual scope is visible.
before you close this tab
Someone in your market is asking AI a question right now. Will your name be in the answer?
The complete diagnostic shows which questions in your industry get your brand cited, which get your competitors cited, and the size of the gap. It is free, with no commitment.
Get the complete diagnosticno meeting to schedule