AI Presence Diagnostic
Measures whether the brand is cited when someone asks about what it sells, on Google, AI Overviews, ChatGPT, Gemini, Perplexity and Bing. Always a percentage with the number of checks next to it.
- The baseline: today's number, with a date
- The absence map: which questions the brand doesn't exist for
- The list of who takes its place
Full audit
Seven topics, eight layers, 56 checks. Every finding comes out with the fix next to it, in order of impact. Runs on the site with nothing to install, and the report is presented by our team or by your agency.
- PDF report with the agency's brand
- Fix list prioritized by weight in the score
- Presentation meeting with the client
What the full audit looks at.
Open each topic: what the audit looks at, how it fixes it, what you get and a real example.
01 · AI presence
Is the brand cited when someone asks about what it sells? The baseline every result is compared against.
Without this one, none of the others mean anything: it is the baseline every later result is compared against.
The question this theme answers: Is your brand cited when someone asks an AI about what you sell?
What the audit looks at
- The real commercial questions of your market, not volume keywords
- Citation across the engines measuring today: Google, AI Overviews, ChatGPT, Gemini, Perplexity and Bing, with the four in activation joining when they switch on
- Presence percentage always with the number of checks beside it
- Who showed up instead of you, question by question
- Breakdown by market, when you serve more than one city
How it fixes, step by step
- Gather the market commercial questions, the ones that decide a purchase, and run them on the engines that are measuring
- Record, per question, the position, the cited snippet and the domains that showed up instead
- Prioritize the questions with no good answer where no competitor dominates: that is where getting in costs least
- Repeat every 15 days and compare against the baseline, never against a one-day snapshot
What you get in this theme
- The baseline Today number, with a date and a denominator, to compare against a month from now.
- The absence map Which questions you do not exist in, and on which engines.
- The list of who holds the ground The competitors cited instead of you and the pages of theirs that were cited.
Example · Villa Zampa, pet shop, Greater São Paulo For the question “best pet shop with grooming near me”, 37.9% presence across 956 checks in a 30-day window on Perplexity, cited first in 117 of them. It is the baseline: without it, any later improvement would be opinion.
What this theme does not do. It does not give the cause. It shows the size of the problem, not the reason. Finding out whether it is a server block, a missing entity or shallow content is the job of audits 02, 04 and 05.
02 · Access
Can the AIs' automated readers get in, read and understand the site?
It is the heaviest layer in the score and the one the market ignores most: a closed door cancels everything after it.
The question this theme answers: Can the automated AI reader get in, read and understand your site?
What the audit looks at
- The server response for each agent, one by one, with the real status code
- robots.txt: what it allows and what it blocks without anyone noticing
- CDN anti-bot and firewall rules that shut out the AI reader
- Whether the content exists in the first response HTML or only after JavaScript
- Time to first byte, which decides whether the agent waits or gives up
How it fixes, step by step
- Allow the AI readers in robots.txt: GPTBot, ClaudeBot, PerplexityBot, Google-Extended and OAI-SearchBot
- Create the exception in the CDN anti-bot and the firewall for the same agents, one by one
- Make sure the main content is in the first response HTML, without depending on JavaScript
- Cut time to first byte: an agent that waits too long gives up before reading
What you get in this theme
- The per-agent test The table with each reader and the status code it received.
- The layer to blame Where the block is: robots, firewall or anti-bot.
- The fix The configuration line to change, in order of impact.
Example · The most common pattern in audits The same page returns 200 to Googlebot and 403 to ClaudeBot. The site exists for Google and does not exist for an entire engine, with a high technical score everywhere else. One exception in the CDN anti-bot fixes it, and nobody sees it without testing agent by agent.
What this theme does not do. It does not touch your server. The audit points it out and explains it; opening the door depends on access to the client CDN or hosting, and that goes into the platform fix queue.
03 · Technical SEO
Can Google crawl, index and rank what the company publishes?
The Google front goes into the same sweep, because half the client budget still lives there.
The question this theme answers: Can Google crawl, index and rank what you publish?
What the audit looks at
- Real indexing: what is in the index and what was discarded, with the reason
- Wrong canonicals, duplicated and pointing to a page that does not exist
- Chained and looping redirects, which burn crawl budget
- Core Web Vitals measured in the field, not in a lab
- Internal link structure and orphan pages
How it fixes, step by step
- Fix canonicals pointing to a nonexistent page or to the wrong page
- Collapse every redirect chain into a single hop, and eliminate the loops
- Link orphan pages into the navigation, or remove them from the index on purpose
- Attack the lowest field Core Web Vital first, which is what the crawler feels
What you get in this theme
- The inventory How many pages exist, how many are indexed and why the others are not.
- The technical queue The errors ranked by how much traffic each one costs.
- Step by step What the dev needs to do, in dev language.
Example · A three-hop redirect The old page sends to a second, which sends to a third, which sends to the final one. Every hop is crawl budget burned, and on part of the visits Google gives up before arriving. The fix is a single hop, from the first to the last.
What this theme does not do. It does not write content. It handles the structure that lets content be found. What to publish is audit 05.
04 · Entity
Does the machine understand what the company is, where it operates and why to trust it?
A language model cites entities, not loose pages. Without a clear entity, you are anonymous text.
The question this theme answers: Does the machine understand what your company is, where it operates and why it should be trusted?
What the audit looks at
- JSON-LD markup connected by identifier, with Organization and Person pointing at each other
- Authorship with a real person, verifiable outside your own site
- Coherence between what the site says and what the rest of the internet says about you
- Presence in the databases models read to confirm you exist
- Address, phone and hours consistent across the whole chain
How it fixes, step by step
- Declare Organization and Person with identifiers, each pointing to the other
- Sign with a real person and link the author to verifiable profiles outside the site
- Keep address, phone and hours identical on the site, the map and every directory
- Publish a canonical source about your own company, which is what the models read to confirm who you are
What you get in this theme
- The current graph What the machine currently understands you to be, with the gaps.
- The new markup The JSON-LD ready to paste, with the identifiers linked.
- The inconsistencies Where the internet disagrees with your site, and what to fix first.
Example · Vynco itself The organization block declares founder and employee linked by identifier to the person Alex Pinhol, with his networks as sameAs. It is the same exam the audit applies to the client, and it would be incoherent to demand it of whoever hires us without doing it here.
What this theme does not do. It does not create authority from nothing. Markup organises what exists; it does not invent reputation. If there is nothing external about the company, that is a job for content and press, not for code.
05 · Content
Do the pages answer the questions that decide the purchase?
It is what AI actually cites: a direct answer, verifiable data and real depth on the subject.
The question this theme answers: Do your pages answer the questions that decide the purchase?
What the audit looks at
- Topic coverage: which funnel questions have a page and which do not
- Cannibalisation between pages competing for the same question
- Whether the answer appears in the first sentence or after three paragraphs of preamble
- Verifiable data instead of adjectives
- Depth per subject compared with whoever is being cited
How it fixes, step by step
- One page per decision question, with no two pages competing for the same answer
- The answer in the first sentence, and the rest of the page backing it
- Verifiable data instead of adjectives: number, deadline, address, condition
- Depth compared with the page being cited instead of yours, theme by theme
What you get in this theme
- The coverage map Each commercial question with the page that answers it, or the gap.
- The editorial queue What to write first, by the size of the gap and the purchase intent.
- What to rewrite The pages that exist but do not answer, with the fix pointed out.
Example · Águas Marek, Santo André For “water truck in Santo André”, the answer cited first with address, phone and 24h emergency service, because the page had that data marked up and the answer in the first sentence. An urgent service: whoever asks calls the first name.
What this theme does not do. It does not publish on its own. The audit says what is missing; the platform content engine is what produces it, and everything passes automatic review before going live.
06 · Competition
Who shows up instead of the brand, and what do those sites have that the brand's doesn't?
Knowing who took the citation is what turns a score into a repair plan.
The question this theme answers: Who shows up instead of you, and what do those sites have that yours does not?
What the audit looks at
- Citation share by question and by engine
- The most cited hosts in your market, with how many times each
- The specific pages that were cited, not just the domain
- What those pages do differently in structure and depth
- Movement over the weeks: who is climbing and who is falling
How it fixes, step by step
- List the hosts cited on each question, with how many times each
- Open the page that won and compare structure, depth and markup with yours
- Copy the method, never the text: what it answers, in what order and with what proof
- Measure again in 15 days to know whether the gap closed
What you get in this theme
- The real podium Who your market cites, with the total number of checks.
- The page that won The cited content, opened up and compared against yours.
- The actionable difference What to copy in method, not in text.
Example · Progrades, São Paulo The same company cited on two different questions, “fence rental Vila Mariana” and “queue barrier rental São Paulo”. Not a lucky query: the entity being understood. Comparing with whoever came first shows what is missing to climb.
What this theme does not do. It does not copy competitors. The comparison exists to find the pattern that works, not to reproduce someone else text, which besides being wrong would not get cited twice.
07 · Local
Is the brand cited in the answer for people asking by city and neighborhood?
Market by market, location by location. That is where most Brazilian companies live.
The question this theme answers: Are you cited in the answer given to someone asking from your city?
What the audit looks at
- Measurement by neighbourhood, city and region, with the same question in every market
- The company map profile: complete data, right category and consistency with the site
- One page per location, with address and hours marked up in the code
- Reviews and what they say, which is a citable source for the model
- Local competitor cited in your place, market by market
How it fixes, step by step
- One page per location, with address, phone and hours marked up in the code
- Map listing with the right category and the same data as the site
- Reply to reviews, which are a citable source for the model
- Measure the same question market by market and fix the market with the most opportunity first
What you get in this theme
- The market by market map Where you are cited and where you are not, city by city.
- The page per location The structure that makes each market exist for the machine.
- The local queue Which market to fix first, by the size of the opportunity.
Example · King Star Colchões, Grajaú For “mattress store Grajaú”, the answer brought the address and opening hours along, which only happens when the location data is marked up in the code. It is the per-location page making the market exist for the machine.
What this theme does not do. It does not replace the map listing. It audits and guides, but keeping the listing, photos and reviews up to date is continuous work for the client or the agency.
Request the audit of your site, or your client's site.
The preview comes out right away with the score and the three most serious issues. The rest goes in the report, and our team gets in touch to present it.
See more
Keep going from here
Other free tools and other content from the site, chosen by the subject of this page.
Start with the measurement, no card
The free tools run right now, without an account. When you want to take this to your client portfolio, the start page shows the ways in, and none of them asks for a card.
See how to start