Why it matters

• ●When a shopper asks an assistant what to buy, the answer usually names a handful of brands. If you're not one of them, you're not in the running.
• ●Many of these answers send no click, so they barely show up in your analytics or Search Console.
• ●An assistant may know your brand when asked about it directly, yet never bring it up on its own. Only unprompted mentions bring new customers.

How to do it in REVVDA

1. 01

   Check that AI assistants can read your site.

   Open Visibility and go to the AI crawlers tab. Every bot that fetches pages for answers (OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot, Bingbot) should show Can read. If one is blocked, REVVDA has already written a ticket to fix it, and this comes first.

2. 02

   Review the buyer questions.

   In Buyer questions, open Manage questions. Keep questions your buyers really ask ("best linen kaftan brands for a beach holiday"), not your brand name. Use Edit questions to add your own.

3. 03

   Run the check.

   Click Run now, or wait for the scheduled run shown under Next run. REVVDA asks every question to ChatGPT, Perplexity, Claude and Gemini and stores every answer.

4. 04

   Read the headline numbers.

   On the Overview tab, Mention rate shows how often you're named unprompted. By assistant splits it per assistant, Sentiment when named shows the tone, and Cited instead most names the brand that wins when you don't.

5. 05

   Start with the fixes that close the gap.

   Click Fixes that close the gap. It opens the Fix queue filtered on the changes most likely to get you named, ranked by estimated gain.

Why it matters

• ●You can't win a recommendation if you don't know who holds it today.
• ●A competitor that ranks below you on Google can still be the one AI recommends, because assistants read and trust pages differently.
• ●The type of site that gets cited (retailer, forum, competitor, magazine, video) tells you what kind of work will change the answer.

How to do it in REVVDA

1. 01

   See who wins most often.

   In Visibility → Overview, the Cited instead most card names the brand AI recommends most when it doesn't name you.

2. 02

   See who wins each question.

   Open Buyer questions. The Recommended instead column shows, question by question, who was named in your place.

3. 03

   See where AI finds its answers.

   Open Sources AI cites. Citations are grouped by type (retailers and marketplaces, forums and communities, competitor brand sites, editorial and guides, video, your site), each with a short What to do.

4. 04

   Turn off-site sources into tasks.

   Under Most-cited pages you don’t own, click Create task on the pages that matter: reply to a forum thread from your brand account, update your product data on a retailer, pitch a magazine. These tasks land in Fix → To fix.

5. 05

   Compare the winning pages with yours.

   For competitor sites, open the matching pages of yours in Explore → Pages. Their pages usually win on facts, so generate content fixes for the gap.

Why it matters

• ●A single unanswered question can repeat thousands of times a month across assistants.
• ●Gaps tell you exactly what content is missing: a fact on a product page, a collection intro, a sizing or care guide.
• ●Working question by question keeps the effort focused on topics that bring buyers.

How to do it in REVVDA

1. 01

   Open the question list sorted by gap.

   Go to Visibility → Buyer questions → Latest results. Questions are sorted by gap, with the biggest misses first.

2. 02

   Read each row.

   For each question, Named you? shows the result for ChatGPT, Perplexity, Claude and Gemini, and Recommended instead shows who won. Questions where you're named first are marked Winning.

3. 03

   Fix gaps your pages can close.

   Click Review fix when REVVDA has already prepared a change, or Generate fix to create one. Fixes go to the Fix queue for approval.

4. 04

   Turn topic gaps into guides.

   When no page of yours answers the question at all, it becomes a guide idea in Content, with the number of buyer questions and the products that fit.

5. 05

   Add the questions you're missing.

   In Manage questions, add questions your sales and support teams hear every week, so the next run covers them.

Why it matters

• ●An outdated price, fabric or return policy repeated by an assistant reaches buyers you never see.
• ●Assistants often describe a brand using retailers, forums and magazines rather than the brand's own site.
• ●A negative or wrong description is easier to fix once you know which source it comes from.

How to do it in REVVDA

1. 01

   Check the tone.

   In Visibility → Overview, Sentiment when named splits the answers that name you into positive, neutral and negative.

2. 02

   Read the answers themselves.

   Open Reports → AI visibility run. It keeps every question and every answer, so you can read exactly how each assistant describes you.

3. 03

   Find where the description comes from.

   In Sources AI cites, see which retailers, forums and magazines assistants rely on. If a retailer shows different product data from your site, update it: AI quotes whichever version it finds.

4. 04

   Answer where buyers ask.

   For forum threads that assistants cite, reply from a clearly named brand account. Never post anonymously or in bulk. Use Create task to track it.

5. 05

   Get alerted.

   In Reports → Automatic sending, keep the Weekly digest on so changes in mentions reach your team every Monday.

Why it matters

• ●One good or bad answer proves nothing; a trend over several weekly runs does.
• ●Linking fixes to results tells you which kind of work to repeat.
• ●Management wants a number that moves, not a list of tasks.

How to do it in REVVDA

1. 01

   Keep the run on a fixed schedule.

   In Visibility → Buyer questions, the Next run line shows the schedule (for example every Monday). Keep the same questions between runs so results stay comparable.

2. 02

   Compare runs.

   Open Run history to see the mention rate run by run, per assistant.

3. 03

   Link changes to results.

   In Activity, Scores over time shows your SEO and GEO scores scan after scan, and the log shows every fix deployed in between, with who approved it.

4. 04

   Follow real visits.

   With Google Analytics 4 connected in Settings, AI referral visits on the Visibility overview counts visits and orders coming from AI assistants over the last 28 days.

5. 05

   Share it.

   Reports → Monthly performance puts scores, fixes shipped and AI visibility in one PDF.

Why it matters

• ●Quotability is one of the four signals in the GEO score, alongside Data, Facts and Semantic structure.
• ●A description like "Classic pajama set. 100% cotton." gives an assistant nothing to say. A precise, checkable sentence gives it a reason to name you.
• ●The same sentences help shoppers decide faster, so this work helps conversion as well as AI visibility.

How to do it in REVVDA

1. 01

   See where you stand.

   In Explore, the Quote score under GEO shows how quotable your pages are, and Scores by page type shows which templates are weakest.

2. 02

   Open the content fixes.

   In Fix, filter on Content. Open Rewrite product descriptions so AI can quote them.

3. 03

   Review the samples.

   For each sample page, REVVDA shows the current description (Now) and the rewrite (Proposed), with every fact numbered.

4. 04

   Check every fact.

   The Fact check panel shows where each fact comes from in your product data. A sentence REVVDA couldn't find is flagged: choose Remove sentence or It’s true, keep it.

5. 05

   Approve and let the scan verify it.

   Click Approve. REVVDA publishes through your store's API and logs every change; the next scan measures the new Quote score. You can undo for 30 days.

Why it matters

• ●Assistants favor pages that state facts clearly over pages that describe a mood.
• ●Your best sellers are the pages buyers ask about most, so they're where proof pays first.
• ●Facts in text can be read by Google and AI; facts in images can't.

How to do it in REVVDA

1. 01

   Find the fix.

   In Fix, filter on GEO and open Add proof blocks to 10 best sellers. It lists the facts it will show: fabric, origin, care and certificate.

2. 02

   Check the dependency.

   Proof blocks build on your product data. If the step shows Has a dependency in Plan, approve the step it depends on first (for example structured data).

3. 03

   Review each block.

   Open Review. Each sample shows the block REVVDA will add, built only from your product data. Any environmental wording is checked by Compliance before it can go live.

4. 04

   Roll out carefully.

   Approve for the first pages, check them, then roll out to the rest. Under Settings → Approval rules, Site-wide fixes start small applies the first 50 pages before the rest.

5. 05

   Extend to the next pages.

   Once verified by the next scan, repeat on the next group of products that buyers ask about in Visibility → Buyer questions.

Why it matters

• ●The Data signal in the GEO score measures structured data coverage, and it's often the lowest score on a catalog.
• ●One template rule can add markup to thousands of product pages at once.
• ●Missing fields (material, color, shipping) leave AI to guess the facts buyers ask about.

How to do it in REVVDA

1. 01

   Check your Data score.

   In Explore, look at Data under the GEO score, then at Scores by page type to see which templates lack markup.

2. 02

   Open the structured data fixes.

   In Fix, filter on GEO. You'll find fixes such as Add Product structured data (JSON-LD), Add Organization structured data, Add BreadcrumbList structured data and Mark up FAQ blocks with FAQPage schema.

3. 03

   Check field coverage.

   In Review, the preview shows the exact markup on a sample page, and the coverage table shows each schema field, the product data it comes from, and the share of pages where it's filled. Warnings list the pages where a field is empty.

4. 04

   Choose the roll-out.

   Pick Start with 50 pages (recommended) or deploy to all pages in batches. On Shopify, the REVVDA app embed writes the markup into each page's head, with no theme edits; turning it off removes it everywhere.

5. 05

   Let the scan verify it.

   The fix lists what the next scan checks, for example one valid Product block per page and prices that match the live page.

Why it matters

• ●AI engines use tags like main, article, h1 and figure to tell the product content from menus and footers.
• ●Template changes can't be pushed through a store's content API, so they need a developer, and a clear ticket.
• ●It's often the single biggest gain on the GEO score.

How to do it in REVVDA

1. 01

   Spot the blocker.

   On Home, the Main blocker card shows the issue holding back the most pages, for example Use semantic HTML in the product template.

2. 02

   Open the ticket.

   Click Open ticket. REVVDA explains why your platform can't apply it automatically, and shows pages affected, effort and estimated gain.

3. 03

   Check the acceptance checks.

   Each ticket lists what the next scan will verify, for example: product content inside one main with an article; exactly one h1 per page holding the product name; product images wrapped in figure.

4. 04

   Hand it to your developer.

   Use the Suggested change, then Copy as ticket or Publish to Notion. Enrich ticket adds evidence and sample pages if your developer needs more context.

5. 05

   Track it to verified.

   In Plan → Board, the ticket moves from Ready for dev to In development, then Shipped, waiting for scan and Verified by scan when every check passes.

Why it matters

• ●Thin descriptions give neither Google nor AI anything to rank or quote.
• ●The facts usually already exist in your product data, care labels and size charts. They just aren't on the page.
• ●Facts shown only in images, such as care instructions, are invisible to search engines and AI.

How to do it in REVVDA

1. 01

   Find the thin pages.

   In Explore → Pages, sort by GEO score: product pages with low Facts scores are the candidates.

2. 02

   Open the related fixes.

   In Fix or Plan, look for content steps such as Rewrite product descriptions so AI can quote them, Add origin and fabric to kids’ products, Add care instructions as text, not images and Add size guide text to dresses.

3. 03

   Generate the rewrite.

   Click Generate fix. REVVDA writes from your product data only and shows the word count before and after.

4. 04

   Check the facts.

   In Review, the Fact check panel shows the source of every fact. Remove anything that isn't in your data, or confirm it's true.

5. 05

   Approve in batches.

   Approve the sample, then the rest of the batch. The next scan measures the change, and you can undo for 30 days.

Why it matters

• ●When no brand page answers a question, assistants cite magazines, forums and competitors.
• ●Guides like sizing, care or fabric comparisons stay useful for years.
• ●A guide that links to your products turns an AI citation into a visit.

How to do it in REVVDA

1. 01

   Pick an idea.

   Open Content. Each Guide idea shows the buyer questions it answers, the monthly search impressions, who is cited instead, and how many of your products fit.

2. 02

   Write the draft.

   Click Write draft. REVVDA writes it from your product data only (fabrics, care, sizes) and links to the matching products.

3. 03

   Edit it.

   In the editor, adjust the text and use Link to product to add products. Environmental wording is flagged as you type.

4. 04

   Clear the checks.

   The Checks panel scores the draft: length, title and meta description length, the topic in the title and first paragraph, at least three sections, a questions section, at least three product links, no environmental claims, and facts taken from your product data.

5. 05

   Send it for approval.

   Click Send for approval. Approving publishes it to your blog with Article and FAQPage markup; later scans flag guides whose products have changed.

Why it matters

• ●Titles and meta descriptions decide what shoppers see in search results before they click.
• ●Duplicate titles across color variants make pages compete with each other.
• ●Empty alt text leaves product images invisible to search and accessibility tools.

How to do it in REVVDA

1. 01

   Filter the SEO fixes.

   In Fix, filter on SEO. Metadata fixes include Write missing meta descriptions, Lengthen short page titles, Shorten titles over 60 characters, Make duplicate titles unique across colour variants and Write alt text for product images.

2. 02

   Review sample changes.

   Each fix shows a few sample pages with the current value (Now) and the proposed one, and where it goes, for example your store's SEO description field.

3. 03

   Edit if needed.

   Use Edit to adjust the wording rule before approving. Skip moves to the next fix.

4. 04

   Approve the batch.

   Click Approve. REVVDA deploys in batches and logs every change in Activity, where you can undo for 30 days.

5. 05

   Handle template issues separately.

   Missing H1s usually come from a template, not page content. They appear under Needs developer with a ticket, for example Add an H1 to collection pages.

Why it matters

• ●Audits often list hundreds of issues with no order, and teams fix the easy ones first.
• ●A template fix or a blocked AI crawler can outweigh hundreds of small page edits.
• ●An estimated gain per fix lets you defend priorities in planning meetings.

How to do it in REVVDA

1. 01

   Start from the biggest gains.

   On Home, Ready now, biggest gains first lists the fixes ready to approve, each with its estimated gain in SEO or GEO points.

2. 02

   See the total.

   In Fix, the If you approve all panel shows the combined estimated effect on your SEO score, GEO score and high-risk claims. The next scan measures the real change.

3. 03

   Clear the main blocker.

   The Main blocker card on Home shows the single issue holding back the most pages, often a template change.

4. 04

   Follow the plan's order.

   In Plan, Why this order explains the ranking: access for AI crawlers first, high-risk claims next, then template fixes before page-by-page work.

5. 05

   Separate quick approvals from developer work.

   Fixes REVVDA can deploy take about two minutes to review. Developer tickets show their effort (15 minutes to a day), so you can plan them into sprints.

Why it matters

• ●A catalog is a handful of templates repeated thousands of times.
• ●Collection pages often score far lower on GEO than product pages, and they're the pages buyers land on from broad questions.
• ●Comparing SEO and GEO side by side shows pages that rank on Google but that AI can't use.

How to do it in REVVDA

1. 01

   Open the table.

   Go to Explore. Scores by page type shows, for each type, the number of pages, the SEO score and its parts (Tech, Struct, Quality), and the GEO score, color-coded from under 40 to 75 and above.

2. 02

   Find the weakest template.

   Look for page types with many pages and a low score, typically product listings (collections) on GEO.

3. 03

   Drill into pages.

   Open Pages to see individual pages with their SEO score, GEO score and open issues. Click a page to see its issues and the fixes for it.

4. 04

   Act on the template.

   Fixes that apply to a whole page type, such as Rewrite collection intros as short, quotable facts, appear in Fix and Plan.

5. 05

   Check what's excluded.

   Excluded URLs lists pages left out of scores, such as internal search results, so they don't distort the averages.

Why it matters

• ●Keywords on positions 8 to 20 with high impressions are the fastest to move.
• ●Search Console shows real demand, not estimates.
• ●The same data shows which topics deserve a guide.

How to do it in REVVDA

1. 01

   Connect Search Console.

   In Settings → Integrations and approvals, connect Google Search Console. It syncs daily: clicks, impressions and positions.

2. 02

   Open the keyword table.

   In Explore → Keywords, each keyword shows its languages, how often it appears on your site, clicks over 28 days, impressions and average position.

3. 03

   Pick the near-wins.

   Sort by impressions and look for keywords with an average position between about 8 and 20. Those pages are close to page one.

4. 04

   Improve the matching pages.

   Open the pages behind those keywords in Explore → Pages and apply their open fixes: titles, descriptions, facts.

5. 05

   Turn demand into guides.

   Topics with impressions but no page that answers them appear as ideas in Content, with their monthly impressions.

Why it matters

• ●A site can expose thousands of internal search URLs that Google crawls instead of your products.
• ●They compete with your real collection pages for the same queries.
• ●They drag down average scores and hide the pages that need work.

How to do it in REVVDA

1. 01

   See how many there are.

   In Explore → Excluded URLs, internal search results are listed and kept out of your scores, so they don't distort them.

2. 02

   Open the fix.

   In Plan, find Keep internal search pages out of the index. It needs a developer because it changes your robots rules, and the ticket shows the pages affected and the estimated gain.

3. 03

   Ship the ticket.

   Click Open ticket, then Copy as ticket or Publish to Notion. The acceptance checks say what the next scan will verify.

4. 04

   Clean up thin tag pages too.

   Remove thin tag pages from the index is a fix REVVDA can deploy after your approval.

5. 05

   Verify.

   After release, the next scan checks the rules and moves the ticket to Verified by scan on the Board.

Why it matters

• ●Each language version competes in its own market and in its own AI answers.
• ●Broken hreflang tags send shoppers to the wrong language version.
• ●Text left in the wrong language on a page confuses both search engines and assistants.

How to do it in REVVDA

1. 01

   Compare coverage by language.

   In Explore → Keywords, the Languages column shows where each keyword appears (for example EN, FR, PT). A keyword missing from one language is a gap.

2. 02

   Scan one market at a time.

   In Settings → Scans and scopes, create a scope such as French product pages so a country manager sees only their pages.

3. 03

   Fix the tags.

   In Plan, look for steps such as Fix hreflang tags in the theme, Add hreflang to Portuguese pages and Add a language switcher with real links.

4. 04

   Fix mixed-language pages.

   Fix French text on English URLs and similar fixes correct pages where text is in the wrong language.

5. 05

   Check AI visibility per market.

   Add buyer questions in each language in Visibility → Manage questions, so each market has its own mention rate.

Why it matters

• ●The rules come from Directive (EU) 2024/825 (ECGT), which amends the Unfair Commercial Practices Directive. They apply to claims already on your site, not only to new ones.
• ●There is no exemption for small businesses and no transition period.
• ●Large catalogs repeat the same claim across hundreds of pages, so one sentence in a template becomes hundreds of problems.

How to do it in REVVDA

1. 01

   Open the claims list.

   Go to Compliance. Claims found counts each unique wording once, with how many times and on how many pages it appears.

2. 02

   Start with the high-risk claims.

   Claims are scored from 0 to 100: High risk is 65 and above, Medium 35 to 64, Low under 35. The table shows the risk, the claim, the pages, the proof found on the page and the status.

3. 03

   Read why each claim is a risk.

   Open a claim to see its category, the Rule it may breach with its Official sources, and where it was found.

4. 04

   Review the proposed fix.

   Click Review fix: REVVDA shows each sample page now and with the fix. Compliance fixes need legal sign-off before they can deploy.

5. 05

   Hand the list to legal.

   Click Review file for legal to export every claim with its pages, rule, proposed fix and sign-off status.

Why it matters

• ●Under the ECGT rules, generic claims like "eco-friendly" or "green" are banned unless you can show recognized excellent environmental performance.
• ●The US FTC Green Guides also warn against unqualified general claims.
• ●Specific facts, such as a material, a share and a certificate, are more convincing to shoppers and easier for AI to quote.

How to do it in REVVDA

1. 01

   List the generic claims.

   In Compliance, look at claims categorized as Generic claim or Sustainable · generic.

2. 02

   Generate a rewrite.

   Open the claim and click Generate fix (or Review fix if one is ready). REVVDA proposes either a specific statement backed by your product data, or removing the claim.

3. 03

   Check the facts.

   Keep only what you can prove with a document. If no proof exists, removing the wording is the safe choice.

4. 04

   Send for legal sign-off.

   Click Send for legal sign-off. Under Settings → Approval rules, Legal sign-off on green claims ensures nothing deploys until your legal reviewer approves.

5. 05

   Deploy and verify.

   Once approved, the change goes live and the next scan updates the claim's status.

Why it matters

• ●The ECGT rules ban sustainability labels that aren't based on a certification scheme or established by public authorities.
• ●"Certified" with no scheme named is one of the riskiest claims on a product page.
• ●A linked certificate turns a claim into proof that shoppers and AI can trust.

How to do it in REVVDA

1. 01

   Find the certification claims.

   In Compliance, the Proof on page column shows, for each claim: None, Not qualified, Named, not linked or Certificate linked.

2. 02

   Fix the implied certifications first.

   Claims categorized as Certification implied ("certified" with no scheme) carry the highest risk. Review the proposed fix: name the scheme, or remove the word.

3. 03

   Add the links.

   For claims marked Named, not linked, steps such as Add certificate links to “Organic cotton” add the certificate link next to the claim.

4. 04

   Get sign-off.

   Send the change for legal sign-off, then approve it.

5. 05

   Keep it current.

   Certificates expire. Later scans re-check each claim, and Reports → New high-risk claim alert warns your team when a new one appears.

Why it matters

• ●Scattered screenshots and emails make reviews slow and decisions hard to trace.
• ●A dated file shows what you knew and what you did, claim by claim.
• ●Marketing and legal work from the same evidence.

How to do it in REVVDA

1. 01

   Export the file.

   In Compliance, click Review file for legal, or go to Reports → Green claims review file. Download it as CSV or PDF.

2. 02

   Know what it contains.

   Each claim with its pages, the rule it may breach, the proposed fix and the sign-off status.

3. 03

   Route sign-off.

   Under Settings → Approval rules, set Legal sign-off on green claims to required, and name your reviewer. Compliance fixes wait for them.

4. 04

   Get alerts.

   In Reports → Automatic sending, turn on New high-risk claim alert for your legal reviewer.

5. 05

   Keep the trail.

   Activity logs every sign-off and every deployed change, and Change log in Reports exports them with before and after.

Why it matters

• ●"Carbon neutral delivery" or "climate neutral product" based on offsets is now on the EU's list of practices banned in all circumstances.
• ●These claims often sit in delivery banners, footers and checkout pages, not only on product pages.
• ●In the US, the FTC Green Guides require offset claims to be backed by competent and reliable evidence.

How to do it in REVVDA

1. 01

   Find them.

   In Compliance, look for claims categorized as Offset-based claim, such as "carbon neutral delivery".

2. 02

   Check where they appear.

   Open the claim to see every page it's on, including templates like delivery banners and footers.

3. 03

   Choose what to say instead.

   Click Generate fix. Usually the safe options are to remove the neutrality wording or to describe what you actually do, without claiming neutrality.

4. 04

   Get legal sign-off.

   Offset wording is a legal call: send the fix for sign-off before it deploys.

5. 05

   Remove it everywhere.

   Once approved, REVVDA removes the wording on every page in scope and the next scan confirms it's gone.

Why it matters

• ●Audits list problems; teams need an order and an owner for each one.
• ●Some fixes depend on others, such as structured data before proof blocks.
• ●A shared plan keeps marketing, developers and legal working on the same list.

How to do it in REVVDA

1. 01

   Open the plan.

   Go to Plan. It's written from your latest scan, and each step is an issue from Fix, so the plan, the Fix queue and the developer board always agree.

2. 02

   Read the priorities.

   Steps are grouped as P1 · Now · This week, P2 · Next · This month and P3 · Later · This quarter, each with the number of open steps and the estimated points.

3. 03

   Understand the order.

   Why this order explains the reasoning, for example: unblock AI crawlers first, remove high-risk claims next, then template fixes before page-by-page work.

4. 04

   See who does what.

   Each step shows who acts and how long it takes: Approver, then REVVDA deploys, Legal, then approver, Developer or Content team. Has a dependency marks steps that wait for another.

5. 05

   Share it.

   Click Publish to Notion, or use Regenerate plan after a new scan. Progress shows the share of steps live or verified.

Why it matters

• ●Traditional audits end as a long PDF that nobody works through.
• ●Developers need a precise scope: which template, what change, how to verify it.
• ●Checks verified by a scan show what has really shipped, not just what was found.

How to do it in REVVDA

1. 01

   Find the developer work.

   In Fix, open Needs developer. These are changes your platform can't apply through its API, such as theme templates, robots rules or CDN settings.

2. 02

   Open a ticket.

   Each ticket shows the pages affected, the effort, the estimated gain, the acceptance checks, and a Suggested change.

3. 03

   Add context if needed.

   Enrich ticket adds evidence and sample pages for your developer.

4. 04

   Send it to your tools.

   Use Copy as ticket or Publish to Notion, then Start when work begins.

5. 05

   Track it to verified.

   On the Plan → Board, tickets move from Ready for dev to In development, Shipped, waiting for scan and Verified by scan once every acceptance check passes.

Why it matters

• ●Scores and tasks mean little to executives without a trend and a cause.
• ●AI visibility is new: leaders need one clear number to follow.
• ●Regular, dated reports build trust in the work and the budget behind it.

How to do it in REVVDA

1. 01

   Use the monthly report.

   In Reports, Monthly performance gives scores, fixes shipped and AI visibility in one PDF.

2. 02

   Send it automatically.

   Under Automatic sending, turn on Monthly performance PDF for your management list and the Weekly digest for your team.

3. 03

   Show the trend.

   Activity → Scores over time shows your SEO and GEO scores scan after scan.

4. 04

   Lead with the summary.

   On Home, the Health score averages SEO and GEO, and the issues bar shows where every issue stands, from To fix to Verified.

5. 05

   Keep it simple for leaders.

   Switch to Simple mode for a manager's view, and Expert for the full detail.

The challenge: a well-ranked site that AI ignores

On paper, everything looked fine. The site is technically sound (80/100), well structured (74/100), and Google understands it. Around a hundred product pages and a hundred blog posts: a solid foundation.

But buying habits are changing. To choose a coffee, more and more shoppers ask an AI for advice instead of comparing ten Google results. So the starting question was simple: when a customer asks an AI for a recommendation, is the brand part of the answer?

That is what GEO (Generative Engine Optimization) is about: being cited in an AI's answer, not just ranked on Google.

The diagnosis: 3 findings

Finding 1: the brand is missing from recommendations

REVVDA asked 9 GPT and Gemini models 3,415 buying questions: "best coffee", "where to buy", "comparison"… The result: 837 brands cited, and never the client's. (How to find which competitors AI recommends instead of you.)

Comparison queries concentrate the citations. That is where the gap is widest.

Finding 2: it is not an image problem

When the brand is named, not a single answer is negative. AI models know it well (93/100) and praise its smoothness and low acidity.

An AI can know a brand very well… and still never recommend it.

The real issue is discoverability: nothing on the web nudges AI models to cite it. This paradox is not an isolated case: we saw it again with a premium sleepwear brand that AI models know at 97/100.

Finding 3: 4 content blockers

1. 01

   No facts (1/15). Product pages describe feelings, not data: no precise origin, no figures.

2. 02

   No quotable sentences (7/20). Not one short, assertive sentence such as "[Brand] roasts 100% Arabica beans for a coffee that is never bitter."

3. 03

   No structured data (5/25). No Article schema on the blog, no breadcrumbs.

4. 04

   Thin pages. 83 pages have fewer than 300 words and lack industry vocabulary (Arabica, roast profile, tasting notes).

On top of that come classic SEO errors: 20 missing or duplicate H1s, 174 title tags to rewrite, 51 orphan pages.

The solution: a two-step action plan

The audit produced 29 actions, ranked by impact and effort. Rather than launching everything at once, the plan moves in two steps.

Step 1: fix the basics every engine reads

Step 2: make the brand quotable by AI

• ●A clear definition on the homepage: "Who is [Brand]?", in two factual sentences.
• ●Facts on every product page: origin, roast level, flavor profile, brewing tips. (How to turn thin product descriptions into fact-rich ones.)
• ●Comparison content, where citations are won.
• ●Marked-up FAQs (FAQPage) on key pages.

Every action becomes a ticket, with the before, the after and the rationale. On a product page, one vague sentence about how the beans look becomes a full paragraph: origins, roast, flavor notes and customer benefit.

Go further: How to turn an SEO/GEO audit into a prioritized roadmap.

Key takeaways

1. 01

   Measure SEO and GEO separately. A site can score 75/100 on one and 0% on the other.

2. 02

   Write facts, not slogans. Figures, origins, precise tips: that is what AI models pick up.

3. 03

   Target comparisons. That is where competitors capture up to 34.7% of citations.

FAQ

The challenge: a premium brand that AI does not recommend

On paper, the site is solid. It is technically sound (81/100), very well structured (85/100) and has more than 1,200 product pages. Google understands it with no trouble.

But to choose pajamas, a baby gift or matching outfits for the whole family, more and more shoppers ask an AI for advice. So the starting question was simple: when a customer asks an AI for a recommendation, is the brand part of the answer?

That is what GEO (Generative Engine Optimization) is about: being cited in an AI's answer, not just ranked on Google.

The diagnosis: 3 findings

Finding 1: the brand is missing from every recommendation

REVVDA asked GPT and Gemini models 7,423 buying questions: "best cotton pajamas", "where to buy", "comparison"… The result: 814 brands cited, and never the client's.

The audit went further, with 3,374 questions targeted by age and gender, such as "which pajamas for a 35-year-old adult?". Once again: zero mentions, whatever the audience. (How to find the buyer questions where your brand is missing.)

Finding 2: it is not an image problem

When the brand is named, not a single answer is negative. AI models know it very well (97/100) and praise its quality, comfort and timeless style. Their only reservations: price and availability.

AI models know the brand almost by heart… yet never think of recommending it.

In the answers analyzed, they often describe it as a classic children's pajama brand, tied to special occasions. A flattering image, but too narrow to be cited on adult queries. (How to monitor what AI agents say about your brand.)

Finding 3: 4 content blockers

1. 01

   No facts (0/15). The copy talks about elegance and softness, without a single figure: fabric weight, composition, cotton origin, care.

2. 02

   No quotable sentences (6/20). Not one sentence such as "[Brand] designs premium cotton pajamas for the whole family."

3. 03

   Little structured data (7/25). No breadcrumbs on 1,414 pages, and Product and Organization schemas still to be deployed.

4. 04

   Off-topic vocabulary. 21% of the site's keywords are washing instructions (wash, dry, machine). On 1,317 pages, industry vocabulary is almost absent (8% coverage).

On top of that come classic SEO errors: 200 pages without a meta description, 199 title tags to rewrite, and promotional pages with vague titles ("Friend", "Giveaway").

These blockers also weigh on Google: we measured it with Search Console data in the case study of a resort fashion brand.

The solution: a two-step action plan

The audit produced 38 actions, including 9 quick wins. Rather than launching everything at once, the plan moves in two steps.

Step 1: fix the basics every engine reads

Step 2: make the brand quotable by AI

• ●A clear definition on the homepage and category pages: who the brand is, who it is for, which fabrics it uses.
• ●Facts on every product page: composition, cotton origin, fit, care. (How to add citation-ready proof blocks to key pages.)
• ●A complete About page: history, founders, craftsmanship. It is the first place an AI learns who a brand is.
• ●Marked-up FAQs (FAQPage) on category pages: "what fabric should I choose for pajamas?", "what is the difference between two types of cotton?".
• ●Content for adults, to move beyond the "children's pajamas" image: women's nightgowns, men's pajama pants, matching family pajamas.

Every action becomes a ticket, with the before, the after and the rationale. On the gift card page, a single sentence becomes three paragraphs: what the card offers, the collections it covers and instant delivery by email.

Go further: How to turn an SEO/GEO audit into a prioritized roadmap.

Key takeaways

1. 01

   Awareness is not enough. A brand can score 97/100 for AI brand knowledge and still be cited zero times.

2. 02

   AI models repeat the image they have of you. If they tie you to a single audience, create content for the others.

3. 03

   Replace emotion with facts. Composition, origin, care: that is what AI models pick up in their answers.

FAQ

The challenge: growing beyond loyal customers

The site ticks every box: 78/100 for SEO, content quality rated 85/100 and more than 2,000 product pages. Customers who already know the brand find it with no trouble.

But a fashion brand does not grow on loyal customers alone. It grows thanks to shoppers searching for "a long dress for vacation" who do not know it yet, on Google or through an AI. The starting question: does the brand exist for those shoppers?

To answer it, the audit combined two sources: classic SEO, using real Google data, and GEO (Generative Engine Optimization), meaning the ability to be cited in AI answers.

The diagnosis: 3 findings

Finding 1: on Google, traffic comes almost entirely from the brand name

The Search Console data (March to June 2026) leaves no doubt. The homepages capture most of the clicks, and their top queries are all variations of the brand name. Product pages, on the other hand, are almost invisible.

In other words, Google knows the brand, but almost never shows it to someone simply looking for a dress.

Finding 2: in AI answers, it is worse: zero citations

REVVDA analyzed 1,350 answers from GPT-5 mini, Claude Sonnet 4.6 and Gemini 2.5 Flash to buying questions. The result: 819 brands cited, and never the client's.

Yet when the brand is named, AI models know it at 99/100 and have nothing bad to say about it (+47). They describe it as chic and relaxed, made for vacations. Their reservations: price, availability and sizing.

What Google cannot see, AI will not recommend.

A telling detail: on queries such as "satin" or "cotton dress", AI models mostly cite… fabric retailers. The words the brand uses are not the words its customers use.

Finding 3: pages that machines cannot read

1. 01

   No usable structured data (0/25). 318 pages do not tell search engines what they contain: product, price, FAQ, breadcrumbs. (How to audit schema.org coverage across a large catalog.)

2. 02

   Weak HTML structure (7/15). The content is not marked up in clear sections that AI models can break down. (How to fix weak semantic HTML once, at template level.)

3. 03

   Few quotable sentences (6/20) and few facts (3/15). A typical product page has 571 words, but not a single measurable fact or trust signal.

4. 04

   SEO basics to fix. Missing or duplicate H1s, missing or overly long meta descriptions, images without alt text, and 20 empty or placeholder pages to clean up.

Lack of facts was also the main blocker we found at an e-commerce coffee roaster missing from 3,415 AI answers.

The solution: prioritize with real Google data

This is the audit's great strength. Actions are not ranked "on instinct", but according to real demand measured in Search Console. You start where there are already impressions to turn into clicks. (How to use Search Console data to choose which keywords to work on.)

Step 1: the quick wins that reach the most searches

Step 2: make the brand readable and quotable

• ●Structured data everywhere: Product, FAQPage, BreadcrumbList and Organization schemas, delivered as ready-to-use code.
• ●Quotable proof blocks on key pages: fabrics, fits, care, in short, factual sentences.
• ●Comparison content to position the brand against the names AI models cite instead.
• ●The customers' vocabulary: "long dress for vacation" rather than fabric names alone.
• ●A reference guide to the long dress, written in the brand's tone of voice and scored 97/100 before publication, with its own FAQ.

What does a fix look like?

Every action becomes a ticket with the before, the after and the rationale. Example on a product page title tag:

• ●Before: "Model name – Color | Collection | Brand".
• ●After: "Model name: black embroidered cotton long dress | Brand".

The product is finally described in the words a shopper would actually type. (How to fix missing meta descriptions, H1s and titles across thousands of pages.)

Key takeaways

1. 01

   A well-known brand is not a findable brand. If your traffic comes from your name, you only reach your loyal customers.

2. 02

   Google and AI fail for the same reasons. Poorly marked-up pages, no facts, none of your customers' vocabulary: SEO and GEO must be worked on together.

3. 03

   Prioritize with your data, not your gut. Start with the pages that already have impressions to convert.

FAQ