Sample audit · Nothing hidden

This is a complete GEO Visibility Audit, published as a sample so you can read every section before you buy. It was generated for Contramar — a venue we have no commercial relationship with — by the same engine that produces your audit. Yours covers your venue, your neighborhood and your competitive set.

MX$9,999 one-time. Ordered from the Visibility Snapshot your free check produces — no second scan, and you can unlock it with a code instead of a card.

Run your free check to order — 2 minutes →What's included

GEO Visibility Audit · Sample

Contramar

Seafood restaurant · Cuauhtémoc, CDMX · Prepared Aug 8, 2026, 1:58 PM

Grade D

42% of the diner questions we tested are answered without Contramar in them. The highest-weighted fix is “Occasion searches — date night, brunch, celebrations, working, groups” — owned by Marketing / PR, days of work.

The short version

Everything in this audit in one page. If you only read one section, read this one.

Overall · grade D

How often AI names you, weighted across every test we ran.

100

Recommendation

Whether you show up when someone asks for the best in your category.

Prominence

How high you land when you are named at all.

100%

Site readability

Share of your pages an AI crawler can read as finished text.

Clear takeaways

1.About 42% of the high-intent searches we tested are being answered without you in them.
2.Your single biggest leak: occasion searches — date night, brunch, celebrations, working, groups — start there.
3.Your core details match across the sources we checked — that consistency is working in your favour.
4.We read 5 pages of your site and found 4 concrete fixes, 1 of them high priority.

What's in this document

01What to fix first, ranked by money at stake
02Your 90-day plan
03Where your details don't match across the internet
04How your website reads to an AI crawler
05Actionable fixes, page by page
06Every question we asked, and what the models said
07The sources AI is citing about your category
08How this compares to last time and to your peers
09Plain-English glossary

What to fix first, ranked by money at stake

Every issue is scored on two things: how much booking intent sits behind the searches it affects, and how badly you're currently missing them. Work top to bottom — the first row is where the most revenue is leaking.

Priority 1 · Occasion searches (date night, brunch, groups)

Occasion searches — date night, brunch, celebrations, working, groups

Impact score

Occasion searches convert nearly as well as direct ones and are far less contested. 83% of 6 occasion searches tested is unclaimed, and each one maps to a page you can publish yourself.

83% of 6 occasion searches tested at risk·Days of work·Owner: Marketing / PR

Priority 2 · Sources the models read before answering

The sources AI leans on for your category — and whether you're on them

Impact score

Models answer from what they can retrieve. 92% of 24 sources the engines cited, fetched and verified omits you entirely, so every answer built from those pages is built without you.

92% of 24 sources the engines cited, fetched and verified at risk·Weeks of work·Owner: DiscoverCDMX

Priority 3 · Position in the list when you are named

Why Mi Compa Chava is named ahead of you

Impact score

You are already in the answer, so the expensive work is done. 17% of 6 answers where you were named at all sits third or lower, where almost no clicks land — this is the cheapest booking lift available.

17% of 6 answers where you were named at all at risk·Weeks of work·Owner: Owner / GM

Priority 4 · What your own site lets a model say about you

Whether your own site is readable and citable by AI crawlers

Impact score

Your own site is the one source you fully control. 50% of 12 machine-readable signals checked on your site is missing, which caps how confidently any engine can describe or recommend you.

50% of 12 machine-readable signals checked on your site at risk·Hours of work·Owner: Web developer

Priority 5 · Direct “best in the neighborhood” searches

Being named at all when someone asks for the best seafood restaurant in Cuauhtémoc

Impact score

These are the searches with the clearest intent to book tonight, and 0% of 6 direct best-of searches tested returns an answer without Contramar in it. Every point recovered here is demand that already exists.

Nothing at risk yet · 6 direct best-of searches tested·Weeks of work·Owner: Marketing / PR

Priority 6 · Position in the list when you are named

Where you sit in the list when you are named

Impact score

Order is not cosmetic. In a chat answer, the first one or two names get the click, the reservation, and the follow-up question. Models order by how confidently and how richly they can describe a venue — depth of available detail, review consensus, and freshness of what's written about you.

17% of 6 answers where you were named at all at risk·Days of work·Owner: Owner / GM

In plain terms: about 42% of the high-intent searches we tested are currently going to someone else. Closing the top three rows above is where that comes back.

Your 90-day plan

The same fixes, sequenced. Each task says who should do it, roughly how long it takes, and how you know it's actually done — so you can hand this to your team as-is.

Phase 1 — Fix what you control

Days 1–14

Make your own site and listing fully machine-readable. Nothing downstream compounds until this is done.

1.
Add Restaurant, PostalAddress, OpeningHoursSpecification, Menu, Reservation / potentialAction structured data
Publish JSON-LD on your homepage and every venue page covering: Restaurant (name, address, telephone, url); PostalAddress (streetAddress, addressLocality, addressRegion, postalCode); OpeningHoursSpecification (dayOfWeek, opens, closes); Menu (hasMenuSection, hasMenuItem, name, description); Reservation / potentialAction (target.urlTemplate, result.@type). Keep the name, category and neighborhood identical to your Google listing.
Done when Google's Rich Results Test and Schema.org validator both parse the markup with no errors.
Owner: Web developer·Days of work
2.
Correct your primary category on Google
Engines inherit “what kind of place is this” from your primary Google type. Set it to the category you want to be recommended for (“Seafood restaurant”) and keep secondary categories tightly related.
Done when Your Google listing's primary category reads as Seafood restaurant and your site markup uses the matching schema type.
Owner: Owner / GM·Hours of work

Phase 2 — Publish and correct

Days 15–45

Claim the occasion questions you are missing from and make your facts identical everywhere they appear.

1.
Publish 5 occasion pages
One page per question you are missing from, answering it directly in the first paragraph: “Best restaurant in Cuauhtémoc, CDMX for a romantic dinner?”; “Where to eat tacos in Cuauhtémoc, Mexico City?”; “Restaurant in Cuauhtémoc, CDMX for a birthday dinner?”; “Best brunch spot in Cuauhtémoc, Mexico City?”; “Restaurant in Cuauhtémoc, CDMX with good vegetarian options?”. Say who it suits, the room, the price band, how to book.
Done when Each page is live, server-rendered, internally linked from the homepage, and answers its question in the opening lines.
Owner: Marketing / PR·Weeks of work·After: p1-schema
2.
Get named on 8 sources the engines already read
These pages are cited when models answer about your category but omit you: google.com, google.com, google.com, google.com, rosetta.com.mx, micompachava.com, google.com, google.com. Submit, claim or pitch an inclusion on each, using the same name, category and neighborhood wording.
Done when Your name appears in the page text of each source, spelled exactly “Contramar”.
Owner: DiscoverCDMX·Weeks of work
3.
Run a name, address, phone and hours consistency sweep
Every directory, delivery platform, booking platform and social profile carries the identical name, address, phone, hours and category. Mismatches split your entity so models cannot merge the evidence into one confident recommendation.
Done when A spot check of ten listings shows identical name, address, phone, hours and category.
Owner: Owner / GM·Days of work
4.
Start a standing review-and-photo routine
A weekly ask at the table, and fresh captioned photos monthly. Review consensus and freshness are two of the three inputs models use to decide confidence.
Done when Review count is rising week over week and the listing has new photos each month.
Owner: Owner / GM·Days of work

Phase 3 — Earn the recommendation

Days 46–90

Get named on the third-party pages engines retrieve from, then measure the movement.

1.
Close the coverage gap against Mi Compa Chava, La Docena, Campobaja
Take the pages that name your competitors and not you — guides, roundups and neighborhood editorial — and earn a placement on each. Match the language those pages use for the category, then ask for inclusion in the next update.
Done when You are named on at least half the sources that currently name a competitor and omit you.
Owner: Marketing / PR·Weeks of work·After: p2-listings
2.
Earn 5–8 independent editorial mentions
Pitch local guides, neighborhood editorial and curated directories. Each mention must state the same three things — “Contramar”, “seafood restaurant”, “Cuauhtémoc” — because that co-occurrence is exactly what a model learns.
Done when Five or more new independent pages name you as a seafood restaurant in Cuauhtémoc.
Owner: Marketing / PR·Weeks of work
3.
Publish a DiscoverCDMX verified profile
A structured, continuously updated profile on a source AI engines already cite for Mexico City hospitality, with your canonical name, category, neighborhood, hours and booking path.
Done when The profile is live and appears in the citation list of a follow-up scan.
Owner: DiscoverCDMX·Days of work
4.
Re-scan and compare on day 90
Run the same 12 questions again. The audit page shows movement per metric and per question against this baseline, so the roadmap is scored on outcomes, not effort.
Done when A new scan exists for this venue and the trend section shows movement against this baseline.
Owner: DiscoverCDMX·Hours of work

Where your details don't match across the internet

AI assistants trust a venue when the same facts appear the same way everywhere. We compared your listing, your own website, and the pages AI cites about your category. Checked your Google listing, 5 pages of your own site, and 23 third-party pages the engines cited.

The version we treat as correct

Name: Contramar
Category: Seafood restaurant
Neighborhood: Cuauhtémoc
Address: Durango 200, Roma Nte., Cuauhtémoc, 06700 Ciudad de México, CDMX, Mexico
Phone: —
Website: http://www.contramar.com.mx/

Category · Google Business Profile

warning

What's there now

seafood_restaurant

What it should say

A primary category matching how you want to be recommended (“Seafood restaurant”)

Google's primary type is what most engines inherit as “what kind of place is this”. Yours reads as “seafood_restaurant”, so you are competing in the wrong set of questions.

Owner: Owner / GM

Address · Your website

Consistent

What's there now

Full street address and city are present

What it should say

Full street address plus Mexico City in readable page text

The address is how an engine ties your site to the Google record and the neighborhood in a question. Text in an image does not count.

Owner: Web developer

Hours on site · Your website

Consistent

What's there now

Hours are readable on the site

What it should say

Hours in readable text, matching the Google listing exactly

When your site and your listing disagree on hours, engines treat both as unreliable and quietly stop citing either.

Owner: Web developer

Booking path · Your website

Consistent

What's there now

A booking or reservation path exists

What it should say

A visible reservation link or phone-to-book path an engine can hand a diner

An engine will only send someone to book if it can see a booking path. Without one you get described but never actioned.

Owner: Web developer

Menu / offer detail · Your website

Consistent

What's there now

Menu or offer wording present in page text

What it should say

Signature dishes or drinks named in readable text, not only in a PDF or image

Named dishes are what a model quotes back to a diner. A PDF menu is invisible to almost every engine.

Owner: Web developer

Narrative content · Your website

warning

What's there now

Little or no about/story content to quote

What it should say

A few paragraphs describing the room, the concept and who it suits

This is the raw material for the sentence a model writes about you. No narrative means no sentence.

Owner: Web developer

Server-rendered content · Your website

Consistent

What's there now

3,245 characters of server-rendered text on the home page

What it should say

Core content present in the HTML without running JavaScript

Most AI crawlers do not execute JavaScript. If your content only appears after scripts run, they read an empty page.

Owner: Web developer

Crawler access · Your website

Consistent

What's there now

No robots.txt, so nothing is blocked

What it should say

robots.txt allowing AI crawlers

Blocking crawlers is a hard stop: nothing downstream in this audit can be fixed while the door is closed.

Owner: Web developer

Titles and meta · Your website

warning

What's there now

Page titles are duplicated across the site

What it should say

A unique title and description per page, each naming the venue and neighborhood

Titles are the strongest short label a retrieval system has for a page. Duplicated titles collapse your pages into one weak signal.

Owner: Web developer

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · rosetta.com.mx

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 and Gemini 3.1 Pro read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · micompachava.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

Gemini 3.1 Pro read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Third-party mention · google.com

warning

What's there now

Does not mention you

What it should say

“Contramar” named in the page text as a seafood restaurant in Cuauhtémoc

GPT-5.6 read this page while answering about your category. Every answer built from it is built without you.

Owner: DiscoverCDMX

Structured data on your website

Structured data is a hidden summary of your venue that machines read directly. Missing entries mean AI has to guess.

Restaurant

Missing — required

This is the type that tells an engine what kind of venue you are. Without it you are a generic web page, not a seafood restaurant in Cuauhtémoc.

Fields to include: name, address, telephone, url, servesCuisine / amenityFeature, priceRange, image

PostalAddress

Missing — required

The machine-readable form of your address. This is what links your site to the Google record and to neighborhood questions.

Fields to include: streetAddress, addressLocality, addressRegion, postalCode, addressCountry

OpeningHoursSpecification

Missing — required

Required to appear in “open now”, “open late” and day-specific answers.

Fields to include: dayOfWeek, opens, closes

Missing — required

Structured menu items are the most quotable thing you own — they let a model recommend a specific dish, not just a name.

Fields to include: hasMenuSection, hasMenuItem, name, description, offers.price

Reservation / potentialAction

Missing — required

Marks your booking path as machine-actionable so an engine can send a diner straight to a reservation.

Fields to include: target.urlTemplate, result.@type

AggregateRating

Missing — recommended

Gives an engine your review consensus in a form it can compare against competitors without leaving your site.

Fields to include: ratingValue, reviewCount, bestRating

ImageObject

Missing — recommended

Captioned, structured images are increasingly used in multimodal answers and rich AI result cards.

Fields to include: contentUrl, caption

FAQPage

Missing — recommended

Question-and-answer markup maps directly onto how diners phrase things to a chatbot, making your page easy to lift verbatim.

Fields to include: mainEntity.name, mainEntity.acceptedAnswer.text

How your website reads to an AI crawler

Measured on 5 pages of www.contramar.com.mx that we fetched and read as an AI crawler would.

Access & renderability

Whether your pages arrive as finished text. AI crawlers read the plain version, not the version a person sees.

100%

Renderable pages · Good

Pages with friction

Nearly every page arrives as finished text, which is exactly what AI crawlers need.

Content density

How much of each page is your own writing versus repeated template copy.

93%

Unique content

Boilerplate

Most of the readable text is specific to each page, which is what gives models something worth quoting.

Trust signals

What tells a model your pages are credible and current.

Pages with an author

Pages with a date

Outbound sources cited

Transparency pages found

Author names, dates and outbound sources are the trust signals models weigh before repeating a claim about you.

Metadata quality

The title and description each page hands to search and AI systems.

Overall quality score2.4/10

Optimal title length (50–60 chars)0%

Pages with a real meta description100%

Descriptions in the 120–160 char sweet spot0%

Action-oriented descriptions0%

Unique titles across pages20%

Titles read best at 50–60 characters. Descriptions should start with a verb — “Reserve…”, “Discover…”, “Taste…” — and describe the page, not the brand.

Summaries & key takeaways

The block AI engines most often lift word for word.

Pages with a summary / TL;DR0%

Pages with key takeaways0%

Why it matters: AI engines prefer content with a clear summary near the top. It gives them a clean block to lift and cite instead of guessing from the whole page.

Quotability

How many self-contained lines exist for a model to repeat.

Definitions

Steps

Statistics

83%

Short sentences

10.1

Blocks / 1k words · High

Definition“Our dishes are subject to availability.”
Definition“Our prices include taxes and are shown in MXN.”
Statistic“de 115 g $115.00 Tostada de callo 1 pza.”

Structured data we recommend adding

LocalBusiness / Restauranthigh priority

No venue schema was detected, so engines can't confirm what kind of place this is, where it is, or when it's open.

Organizationmedium priority

Organization schema ties your brand name, logo and social profiles into one record models can trust.

Menumedium priority

Menu wording is present but not marked up, so dishes and prices can't be quoted precisely.

BreadcrumbListlow priority

Multi-page site. Breadcrumbs help engines understand how your pages relate.

AggregateRatinglow priority

Ratings on your own site give models a first-party quality signal to cite alongside third-party reviews.

Actionable fixes, page by page

Each fix names the exact pages affected, what to change, and what counts as finished — so it can be handed straight to whoever maintains your site.

01high prioritySchema

Add valid structured data to the affected pages

Structured data lets answer engines interpret what a page is about without guessing from the layout.

Estimated score impact

+1 to +3 overall · +1 to +3 on site readiness

Owners

Web developer

What to change

•Add a JSON-LD block to each affected page that matches what the page actually shows.
•Start with LocalBusiness / Restaurant, then FAQ, Menu and Article types where they apply.
•Validate the markup before publishing so a syntax error doesn't void the whole block.

What counts as fixed

✓Each affected page contains at least one valid JSON-LD block.
✓The schema type matches the visible content on that page.
✓The structured data parses without validation errors.

Current state

5 pages carry no structured data.

Target: Every page exposes valid schema that matches its visible content.

How to confirm the fix

Re-run the audit after publishing. This item clears once valid schema is detected on the affected pages.

Affected pages

//index.html/vinos.html/bebidascontramar.html/english.html

5 affected pages out of 5 pages we read.

02medium priorityTrust

Add author attribution and content dates

Named authors and visible dates are how models decide whether a page is current and credible enough to quote.

Estimated score impact

+1 to +2 overall · +1 to +3 on citations

Owners

Marketing / PR · Web developer

What to change

•Add a byline to editorial pages — “By Chef …”, linked to a short bio.
•Show “Published on” or “Last updated” dates on menu, events and story pages.
•Expose the same values in schema as datePublished / dateModified.

What counts as fixed

✓Editorial and menu pages carry a visible author or team byline.
✓Every informational page shows a date.
✓Schema includes datePublished or dateModified.

Current state

Author attribution is on 0% of pages and dates on 0%. We found 5 outbound authoritative sources.

Target: Bylines and dates on every informational page.

How to confirm the fix

Re-run the audit; both E-E-A-T percentages should rise above 60%.

Affected pages

//index.html/vinos.html/bebidascontramar.html/english.html

5 affected pages out of 5 pages we read.

03medium priorityStructure

Add short summaries to the top of key pages

A two-line summary at the top of a page is the block an answer engine most often lifts verbatim.

Estimated score impact

+1 to +2 overall · +1 to +3 on site readiness

Owners

Marketing / PR · Web developer

What to change

•Add a two- or three-sentence summary at the top of each major page.
•Follow it with three to five bulleted key facts — cuisine, neighborhood, price band, hours, booking.
•Write it in direct language: “Boca Grande is a …”, not slogans.

What counts as fixed

✓Every major page opens with a summary of at least 25 words.
✓The summary states the venue name, category and neighborhood.
✓Key facts appear as a short list rather than buried in prose.

Current state

0% of pages have a summary block and 0% have key takeaways.

Target: Summaries on every major landing and menu page.

How to confirm the fix

Re-run the audit; the summary detection panel should read above 60%.

Affected pages

//index.html/vinos.html/bebidascontramar.html/english.html

5 affected pages out of 5 pages we read.

04medium priorityStructure

Improve title and meta-description quality

Metadata reinforces what a page is about and shapes how it is summarised in AI-assisted results.

Estimated score impact

+1 to +2 overall · +1 to +2 on site readiness

Owners

Marketing / PR · Web developer

What to change

•Rewrite thin or duplicated titles so each is specific to its page, at 50–60 characters.
•Give every page a 120–160 character description that starts with a verb.
•Include the venue name plus the neighborhood in the title.

What counts as fixed

✓Titles are specific, unique and not overstuffed.
✓Descriptions are present and meaningfully describe the page.
✓The metadata quality score reaches at least 7/10.

Current state

Metadata quality is 2.4/10; 0 pages sit below the threshold.

Target: Metadata quality of 7/10 or better.

How to confirm the fix

Re-run the audit after updating metadata; the meta quality score should clear the threshold.

Why your score is what it is — and how to fix it

6 things we checked, in the order that affects bookings most. Each one starts with a plain-English line, then the evidence, then what to do.

Priority 1

Occasion searches — date night, brunch, celebrations, working, groups

Critical

In plain terms: people don't only search by food — they search by plan. Date night, brunch, birthday, a table for ten. If nothing published about you uses those words, you're never offered.

Named in 1 of 6 occasion searches

“Best restaurant in Cuauhtémoc, CDMX for a romantic dinner?” — you were not named. Named ahead of you: Rosetta, Huset, Máximo Bistrot, Havre 77. GPT-5.6 said: “Rosetta is my first choice for a romantic dinner; Huset is especially appealing if you want a leafy courtyard.”
“Where to eat tacos in Cuauhtémoc, Mexico City?” — you were not named. Named ahead of you: Los Cocuyos, Taquería El Califa de León, Tacos Hola El Güero, Taquería Orinoco. GPT-5.6 said: “Try Los Cocuyos for suadero and offal, or El Califa de León for its famously spare steak tacos.”
“Restaurant in Cuauhtémoc, CDMX for a birthday dinner?” — you were not named. Named ahead of you: Ling Ling Mexico City, Blanco Colima, Botánico, Máximo Bistrot. GPT-5.6 said: “Ling Ling works for a celebratory birthday with views, while Blanco Colima or Botánico offer a more food-focused evening.”

What's happening

Occasion questions convert far better than generic ones — the person has already decided what they want to do. Models answer them by matching the occasion words in the question against occasion words published about the venue. If nothing about you says “date night”, “brunch”, or “group dinner”, you are not a candidate no matter how good you are.

What to do about it

Publish one focused page or section per occasion you actually serve, using the diner's words in the heading and the first sentence (for example “Date night in Cuauhtémoc”, “Sunday brunch”, “Private dining for groups of 8–20”). Include what's served, timing, and how to book. Then get those same phrases into your listings and any guide that features you.

Days of work·5 unserved occasion searches in Cuauhtémoc, each with a diner who has already decided to go out.

Priority 2

Whether your own site is readable and citable by AI crawlers

Weak

In plain terms: your own website is the one source you fully control, and right now it's harder for AI to read than it should be. This is the part you can fix this week without anyone's permission.

50/100 across 5 pages we read

Structured data (JSON-LD) — No valid JSON-LD found on any page we read
Venue-specific schema type — No Restaurant/Hotel/Bar/LocalBusiness schema — engines can't confirm what kind of place this is
Story / about content — Little or no about/story content to quote
Published or updated dates — No dates anywhere — content can read as stale to time-sensitive queries
Unique title and meta description — Page titles are duplicated across the site
Headings structure (H1/H2) — Missing a clear H1/H2 heading structure

What's happening

Your site is the only source you fully control, and it's the one engines check to confirm what other pages claim about you. If it doesn't render without JavaScript, has no venue schema, or hides menu, hours, and booking behind scripts, the model can't verify you and falls back to whoever it can.

What to do about it

Fix in this order: structured data (json-ld), then venue-specific schema type, then story / about content, then published or updated dates, then unique title and meta description. Concretely: server-render your key pages as plain HTML, add Restaurant JSON-LD with address, hours, price range and a menu link, and put a real booking URL in the markup.

Hours of work·This is the only part of the picture you can fix this week without anyone else's permission.

Priority 3

The sources AI leans on for your category — and whether you're on them

Critical

In plain terms: AI doesn't invent recommendations. It reads a handful of web pages first. These are those pages — and whether your name appears on them.

You're named on 1 of 23 verified sources

google.com — resolves, does not mention you (https://www.google.com/maps/search/?api=1&query=Mi%20Compa%20Chava%20Roma%20Mexico%20City)
google.com — resolves, does not mention you (https://www.google.com/maps/search/?api=1&query=Campobaja%20Mexico%20City)
google.com — resolves, does not mention you (https://www.google.com/maps/search/?api=1&query=Contramar%20Mexico%20City)
google.com — resolves, does not mention you (https://www.google.com/maps/search/?api=1&query=La%20Docena%20Roma%20Mexico%20City)
rosetta.com.mx — resolves, does not mention you (https://rosetta.com.mx/)
contramar.com.mx — resolves and names you (https://contramar.com.mx/)

What's happening

This is the layer underneath the score. Answer engines assemble recommendations from pages they can retrieve and trust. A page that ranks for your category and doesn't mention you is a recommendation you cannot win — and every competitor listed on it gets named instead of you.

What to do about it

Work the list above in order of how often it was cited: pitch an inclusion or update to each source, verify your details where you already appear, and correct any outdated listing. Where a source is a directory or guide you can join, join it. Where it's editorial, give the writer a reason and an asset to use.

Weeks of work·22 cited sources currently feed answers about your category without you in them.

Priority 4

Why Mi Compa Chava is named ahead of you

Weak

In plain terms: Mi Compa Chava gets named instead of you. Usually not because it's better — because more places on the internet describe it clearly, in the words customers use.

Mi Compa Chava appeared 10 times, average position 1.8

Named for “Where do locals go for a seafood restaurant in Cuauhtémoc, CDMX?” — you were not mentioned at all

What's happening

Models converge on the venue they can describe most confidently for a given intent. A competitor that appears this often is being described consistently, in more places, and in the language of the question — not necessarily because the product is better.

What to do about it

Take their searches one at a time: for each question they win and you don't, publish the page that answers it about you, and get one credible third-party mention that uses the same words. Then re-scan to confirm movement.

Weeks of work·Each of these searches is demand in your neighborhood, in your category, currently routed to someone else.

Priority 5

Being named at all when someone asks for the best seafood restaurant in Cuauhtémoc

Strong

In plain terms: when a customer asks an AI assistant for the best seafood restaurant in Cuauhtémoc, your name has to come up. If it doesn't, the AI simply doesn't know enough about you yet to feel safe recommending you.

Named in 6 of 6 direct best-of searches

“Who is the best seafood restaurant in Cuauhtémoc, Mexico City?” — you were named #1. GPT-5.6 said: “Contramar is my top overall pick; reserve ahead and order the tuna tostadas and pescado a la talla.”
“Can you recommend a highly rated seafood restaurant near Cuauhtémoc, CDMX?” — you were named #1. GPT-5.6 said: “For consistently excellent seafood near central Cuauhtémoc, start with Contramar or the livelier Mi Compa Chava.”
“I'm visiting Cuauhtémoc in Mexico City — which seafood restaurant should I book?” — you were named #1. GPT-5.6 said: “Book Contramar for the classic Mexico City seafood experience; choose Mi Compa Chava if you prefer a louder, more casual meal.”

What's happening

Answer engines already resolve Contramar as a real, recommendable seafood restaurant in Cuauhtémoc. That happens when your name appears, consistently spelled, across the sources these models were trained on and retrieve from.

What to do about it

Protect it: keep your name, category, and neighborhood identical everywhere (site, Google, listings, guides), and keep earning fresh mentions so the association doesn't decay.

Weeks of work·Already earning the recommendation on the highest-intent searches.

Priority 6

Where you sit in the list when you are named

Strong

In plain terms: being mentioned isn't the same as being chosen. In a chat answer the first one or two names get the booking, so third place is close to invisible.

Usually a top-two pick

“Where do locals go for a seafood restaurant in Cuauhtémoc, CDMX?” — you were named #4. Named ahead of you: Mi Compa Chava, El Pescadito, Campobaja, Con Vista al Mar. GPT-5.6 said: “Locals commonly choose Mi Compa Chava for mariscos and El Pescadito for a quick, casual seafood-taco lunch.”

What's happening

What to do about it

Give models more to say about you than your competitors have: a specific, current description of the room, the signature dishes or drinks, the occasion you're best for, price band, hours, and booking path — published on your site with structured data, and mirrored in your listings. Specific detail is what promotes a name from “also mentioned” to “start here”.

Days of work·Moving from mid-list to top-two on the searches you already appear in is the fastest booking lift available to you.

Appendix

Evidence behind every number above

Nothing here appears in the free Visibility Snapshot. This is the raw record the audit was built from: every question asked, each model's verbatim answer, every cited source we fetched and verified, the full competitive set, your site crawl, and movement since your last scan.

Every question we tested

Each engine's position for you, with the answer it gave, captured verbatim during the scan.

Best-of

“Who is the best seafood restaurant in Cuauhtémoc, Mexico City?”

Best position #1

GPT-5.6: named you #1 — “Contramar is my top overall pick; reserve ahead and order the tuna tostadas and pescado a la talla.”

Gemini 3.1 Pro: named you #1 — “Contramar is the absolute best for seafood, but Mi Compa Chava is also incredible.”

Best-of

“Can you recommend a highly rated seafood restaurant near Cuauhtémoc, CDMX?”

Best position #1

GPT-5.6: named you #1 — “For consistently excellent seafood near central Cuauhtémoc, start with Contramar or the livelier Mi Compa Chava.”

Gemini 3.1 Pro: named you #1 — “I highly recommend Contramar or Campobaja for top-rated seafood in the area.”

Best-of

“I'm visiting Cuauhtémoc in Mexico City — which seafood restaurant should I book?”

Best position #1

GPT-5.6: named you #1 — “Book Contramar for the classic Mexico City seafood experience; choose Mi Compa Chava if you prefer a louder, more casual meal.”

Gemini 3.1 Pro: named you #1 — “You must book Contramar well in advance, it is an iconic Mexico City experience.”

Best-of

“Top seafood restaurant options in Cuauhtémoc, Mexico City?”

Best position #1

GPT-5.6: named you #1 — “These are my strongest seafood options, ranging from polished destination dining to casual Baja-style tacos and tostadas.”

Gemini 3.1 Pro: named you #1 — “These are the top seafood spots in the borough, ranging from iconic to casual.”

Best-of

“Where do locals go for a seafood restaurant in Cuauhtémoc, CDMX?”

Best position #4

GPT-5.6: named you #4 — “Locals commonly choose Mi Compa Chava for mariscos and El Pescadito for a quick, casual seafood-taco lunch.”

Gemini 3.1 Pro: did not name you — “Locals love Mi Compa Chava for Sinaloan-style seafood and a great, lively atmosphere.”

Named ahead of you: Mi Compa Chava, El Pescadito, Campobaja, Con Vista al Mar

Best-of

“Best reviewed seafood restaurant in Cuauhtémoc, Mexico City?”

Best position #1

GPT-5.6: named you #1 — “Contramar is the safest consensus choice for acclaimed seafood, though Mi Compa Chava is especially popular for a casual outing.”

Gemini 3.1 Pro: named you #1 — “Contramar consistently gets the best reviews for its tuna tostadas and whole grilled fish.”

Occasion

“Best restaurant in Cuauhtémoc, CDMX for a romantic dinner?”

Not named

GPT-5.6: did not name you — “Rosetta is my first choice for a romantic dinner; Huset is especially appealing if you want a leafy courtyard.”

Gemini 3.1 Pro: did not name you — “Rosetta offers a stunning, romantic atmosphere in an old mansion with incredible cuisine.”

Named ahead of you: Rosetta, Huset, Máximo Bistrot, Havre 77

Occasion

“Where to eat tacos in Cuauhtémoc, Mexico City?”

Not named

GPT-5.6: did not name you — “Try Los Cocuyos for suadero and offal, or El Califa de León for its famously spare steak tacos.”

Gemini 3.1 Pro: did not name you — “Orinoco is great for late-night, while Los Cocuyos in the Centro is legendary.”

Named ahead of you: Los Cocuyos, Taquería El Califa de León, Tacos Hola El Güero, Taquería Orinoco

Occasion

“Restaurant in Cuauhtémoc, CDMX for a birthday dinner?”

Not named

GPT-5.6: did not name you — “Ling Ling works for a celebratory birthday with views, while Blanco Colima or Botánico offer a more food-focused evening.”

Gemini 3.1 Pro: did not name you — “Blanco Colima is gorgeous and has a great upscale vibe for celebrating a birthday.”

Named ahead of you: Ling Ling Mexico City, Blanco Colima, Botánico, Máximo Bistrot

Occasion

“Best brunch spot in Cuauhtémoc, Mexico City?”

Not named

GPT-5.6: did not name you — “Lalo! is my top lively brunch choice; Niddo and Café Nin are better for a smaller, calmer breakfast.”

Gemini 3.1 Pro: did not name you — “Niddo and Lardo are fantastic for brunch, just be prepared to wait for a table.”

Named ahead of you: Lalo!, Niddo, Café Nin, Maque

Occasion

“Restaurant in Cuauhtémoc, CDMX with good vegetarian options?”

Not named

GPT-5.6: did not name you — “Plantasia is the strongest sit-down choice, while Por Siempre Vegana is ideal when you specifically want plant-based tacos.”

Gemini 3.1 Pro: did not name you — “Los Loosers and Por Siempre Vegana offer amazing plant-based takes on traditional Mexican food.”

Named ahead of you: Plantasia, Ojo de Maíz, Por Siempre Vegana Taquería, Los Loosers

Occasion

“Where do locals eat in Cuauhtémoc, Mexico City on a Sunday?”

Best position #3

GPT-5.6: named you #3 — “On Sundays, locals linger over breakfast at El Cardenal or Maque, then favor long lunches at places like Contramar.”

Gemini 3.1 Pro: named you #3 — “Sundays are for barbacoa at El Hidalguense or a long, boozy lunch at Contramar.”

Named ahead of you: El Cardenal, Maque, El Hidalguense, Mercado de San Juan

The web pages AI actually read

These are the pages the AI leaned on when it answered — its (A web page the AI actually read and pointed to when it answered — its source for what it said.). Green means your name is on the page. Red means the page shaped a recommendation about your category without you in it.

We fetched 24 cited sources: 23 resolved, 0 didn't exist, 1 couldn't be read. You are named on 1 of the 23 that resolved. Live web-grounded citations weren't available on this run.

Sources fetched

Mention you

Failed verification

contramar.com.mx

Verified

Contramar

Names you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

rosetta.com.mx

Verified

Rosetta

Does not mention you

micompachava.com

Verified

mayo 2018

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

google.com

Verified

Google Maps

Does not mention you

restauranteelcardenal.com

Verified

Inicio | El Cardenal

Does not mention you

contramar.com.mx

Unreadable

https://www.contramar.com.mx/

Cited by Gemini 3.1 Pro

Who AI names instead — full set

These venues were named in answers where your venue was not mentioned at all. This is not a head-to-head ranking — it's the questions you're missing from entirely.

1Mi Compa ChavaNamed in 10 of the questions we asked · usually around position #1.8

Questions where they appeared and you didn't

Named for “Where do locals go for a seafood restaurant in Cuauhtémoc, CDMX?” — you were not mentioned

2La DocenaNamed in 8 of the questions we asked · usually around position #3.1

3CampobajaNamed in 8 of the questions we asked · usually around position #3.3

4Blanco ColimaNamed in 3 of the questions we asked · usually around position #2.7

Questions where they appeared and you didn't

Named for “Best restaurant in Cuauhtémoc, CDMX for a romantic dinner?” — you were not mentioned
Named for “Restaurant in Cuauhtémoc, CDMX for a birthday dinner?” — you were not mentioned

5Lalo!Named in 3 of the questions we asked · usually around position #3

Questions where they appeared and you didn't

Named for “Best brunch spot in Cuauhtémoc, Mexico City?” — you were not mentioned

6HusetNamed in 3 of the questions we asked · usually around position #3.3

Questions where they appeared and you didn't

Named for “Best restaurant in Cuauhtémoc, CDMX for a romantic dinner?” — you were not mentioned
Named for “Restaurant in Cuauhtémoc, CDMX for a birthday dinner?” — you were not mentioned

Site audit

Read 5 pages on www.contramar.com.mx. Missing or not machine-readable: Structured data (JSON-LD), Venue-specific schema type, Story / about content, Published or updated dates, Unique title and meta description, Headings structure (H1/H2).

Structured data (JSON-LD)Missing

No valid JSON-LD found on any page we read

Venue-specific schema typeMissing

No Restaurant/Hotel/Bar/LocalBusiness schema — engines can't confirm what kind of place this is

Menu or offer detail in textPresent

Menu or offer wording present in page text

Opening hours in textPresent

Hours are readable on the site

Reservation or booking pathPresent

A booking or reservation path exists

Address and cityPresent

Full street address and city are present

Story / about contentMissing

Little or no about/story content to quote

Published or updated datesMissing

No dates anywhere — content can read as stale to time-sensitive queries

Unique title and meta descriptionMissing

Page titles are duplicated across the site

Content renders without JavaScriptPresent

3,245 characters of server-rendered text on the home page

Headings structure (H1/H2)Missing

Missing a clear H1/H2 heading structure

Crawlers allowed (robots.txt)Present

No robots.txt, so nothing is blocked

Pages read

http://www.contramar.com.mx/ — HTTP 200 — Contramar
http://www.contramar.com.mx/index.html — HTTP 200 — Contramar
http://www.contramar.com.mx/vinos.html — HTTP 200 — Contramar
http://www.contramar.com.mx/bebidascontramar.html — HTTP 200 — Contramar
http://www.contramar.com.mx/english.html — HTTP 200 — Contramar

What changed since last time

This is your first scan, so it becomes the starting line. Run it again after you make changes and every number below will show whether it went up or down.

Overall score

62First scan

Recommendation

100First scan

Prominence

89First scan

Occasion coverage

17First scan

Site readiness

50First scan

Citation presence

4First scan

GPT-5.6 recall

100First scan

Gemini 3.1 Pro recall

83First scan

How you compare to similar venues

You're one of the first seafood restaurants scanned in Cuauhtémoc, so there's no peer benchmark yet. It fills in as more venues in your category run the check.

◇ Plain-English glossaryNew to any of these words? Start here

GEO

Generative Engine Optimization — the work of getting AI assistants to recommend your venue, the way SEO got you into Google results.

Answer engine

Any tool that answers a question directly instead of showing a list of links — ChatGPT, Gemini, Perplexity.

LLM

Large language model: the AI behind assistants like ChatGPT that writes the recommendation a customer reads.

Citation

A web page the AI actually read and pointed to when it answered — its source for what it said.

Retrieval

The step where the AI goes and fetches pages before answering. If nothing about you is fetchable, you can't be named.

Structured data

A small block of hidden code on your website that states your address, hours, price range and menu in a format machines read without guessing.

Schema / JSON-LD

The standard format that structured data is written in. Adding it is a one-time job for whoever built your site.

Entity

The AI's internal record of your venue — one name, one category, one address. Inconsistent listings split it in two and weaken both.

Crawler

The automated reader an AI company sends to your site. It sees only what loads as plain text, not what a person sees.

Server-rendered

Your pages arrive as finished text rather than being drawn by scripts in the browser. Crawlers can only reliably read the finished version.

robots.txt

A one-line file on your site that tells AI crawlers whether they're allowed in. Blocking them makes you invisible.

Best-of question

"Best tacos in Roma Norte" — a customer asking for a shortlist. The highest-intent question you can win.

Occasion question

"Where to take a date", "Sunday brunch", "dinner for ten" — searching by plan rather than by cuisine.

Prominence

Not whether you're mentioned, but how early. First two names in a chat answer get almost all the bookings.

Hallucinated source

A page the AI cited that doesn't actually exist — a sign it had very little real information to work with.