A top-ranked national spa franchise scores 38.7 on AI visibility. What other brands can take from it.

By Philipp Groubii · Founder, Pondral · Published May 16, 2026

A national spa franchise has locations across the United States, a top ranking in its category from a national business magazine, and a large base of strong customer reviews.

By every traditional measure, this brand is winning.

So we ran its flagship location through an AI visibility audit. We tested five queries across ChatGPT, Claude, Gemini, and Grok. The kind of questions a real consumer would type: “best luxury day spa near me,” “where should I go for a spa day,” “compare day spas in my area.”

The reason is the page, not the brand.

The brand is strong. The page is weak.

Most businesses assume a simple equation: strong brand + great reviews = AI engines recommend us. That assumption is wrong, and this audit shows exactly where it breaks down.

When we tested a branded query (“Is [this spa] worth it?”), the score jumped to 86 out of 100. AI engines know the brand exists. They can describe it accurately. They'll recommend it when you ask about it by name.

But nobody discovers a new spa by typing the spa's name into ChatGPT. They type “best day spa near me” or “where should I go for a facial in [city].” On those discovery queries, the franchise scored between 21.8 and 49.5.

That gap is the finding: strong on branded queries, weak on discovery. The brand gets into the candidate set that AI engines consider. But when those engines decide which businesses to actually cite in their answer, the answers we read cited smaller local competitors with more detailed pages instead.

Who's cited instead

The local competitors cited ahead of this national franchise in AI responses share three traits. None of them involve brand size or review counts.

Citable uniqueness. One independent spa operates out of a restored historic cottage and lists several local “best of” awards on its page. These are specific, verifiable facts an engine can quote in a response. The franchise location's page reads like a template because it is one, copied across every location with only a short passage changed per city.

Service-specific depth. A competing spa has individual pages for each treatment type: one for facials, one for massage, one for couples experiences. Each page answers the specific questions someone would ask about that service. The franchise has a single location page that lists everything in a promotional bullet list, without prices, without details, without the structured depth a specific query asks about.

Structured data. Several competitors have LocalBusiness schema markup telling AI engines their hours, services, price ranges, and ratings in a machine-readable format. The franchise location page has zero structured data. No JSON-LD of any type. No FAQPage schema, despite having six FAQ questions on the page. No AggregateRating schema, despite having many reviews on Yelp.

Engines quote what they can extract cleanly, and this page gives them little to extract.

The franchise template problem

This is where the audit gets instructive for any business running multiple locations, product pages, or regional variations of the same content.

Each franchise location page we read shares most of its content with every other location page. The rest, a short passage, is location-specific: a mention of a nearby beach, a reference to a local museum, the street address. None of them stands out as the local expert on “day spa in [specific city].”

The result is counterintuitive. In the answers we read, a single-location independent spa with a long page of unique, locally relevant content was cited ahead of a franchise with a nationally recognized brand. The independent spa's page gives an engine more specific material for that local query.

In this audit, the brand was enough to be described well on a branded query. On discovery queries, the better-built local pages were the ones cited.

A simplified model of how AI engines select sources

Engines do not publish how they choose which businesses to cite. This is a simplified model we use to reason about it, in five stages. It is not a documented pipeline.

Stage 1: Query decomposition. The AI breaks “best luxury day spa near me” into sub-questions. What makes a spa “luxury”? What spas exist in this area? What do reviews say about each? What services do they offer?

Stage 2: Candidate retrieval. The engine pulls candidate pages from its index. This uses traditional search signals. We assume the franchise page makes this cut, given how well engines describe the brand on branded queries.

Stage 3: Content extraction. This is where we think the franchise falls short. The AI tries to extract structured information from each candidate page. Pages with clear headers, direct answers, schema markup, and well-organized content are easier to parse. Template-based promotional copy with no structured data gives it less to extract.

Stage 4: Source scoring. In this model, each candidate is weighed on signals such as factual consistency, content structure clarity, domain authority, recency, and specificity. In our reading, the franchise page looks strong on domain authority and weak on structure and specificity.

Stage 5: Citation decision. The AI generates its answer and decides which sources to cite. Engines tend to quote what they can extract cleanly. In this audit, local independents with detailed, structured pages were cited ahead of a national brand with a thin template page.

In this model, the franchise passes Stage 2 and falls short at Stage 3, content extraction, which is the stage structured data and unique content are written for.

The five gaps (scored)

We scored the franchise location page across five components of AI Visibility readiness, each on a 0-to-10 scale.

Total on an earlier readiness checklist: 3 out of 10.

This is an earlier readiness checklist, not the AI Visibility Score. The two use different scales and should not be compared. The franchise's AI Visibility Score in this audit is 38.7 out of 100.

What the first changes look like

The most striking part of this audit is how small the first changes are.

Adding FAQPage schema to the six existing FAQ questions. The content is already written. It just needs to be wrapped in structured data so AI engines can parse it as explicit question-and-answer pairs instead of generic page text.

Adding LocalBusiness JSON-LD with hours, services, and geo coordinates. One block of structured data that states what this business is, where it is, when it's open, and what it offers, in a format an engine can read. The franchise doesn't have it on any of its location pages.

Writing unique local content. Replace the franchise template “about” section with real, locally specific content. Who are the therapists? What makes this location different? What do locals love about it? This puts specific, quotable local facts on a page that today reads like every other location.

We have not re-run this audit after changes like these, so we are not going to put a projected score on them.

What this means for your brand

You're probably not running a national spa franchise. But the pattern in this audit applies to any brand with multiple similar pages, whether those are product variants, regional landing pages, or service categories.

AI cites pages, not brands.

Your brand reputation, your review score, your market position. They get you into the candidate set that AI engines evaluate. That's Stage 2 of the model above. The citation decision comes later, at Stage 3 and Stage 5, where an engine has to extract something from your page and choose what to quote.

If your pages are thin, templated, or missing structured data, you are where this franchise was: in the answers we read, smaller competitors with better-built pages were cited instead.

Frequently asked questions

Does a high Google ranking mean high AI visibility?

No. Strong SEO can help a page get retrieved as a candidate (Stage 2 of the model above), but it doesn't guarantee citation. In this audit, a brand with a strong search presence still scored between 21.8 and 49.5 on the discovery queries we tested.

How important is schema markup for AI visibility?

Schema markup is cheap to add and does one specific thing: JSON-LD structured data (LocalBusiness, FAQPage, Product, Article) states a page's facts in a machine-readable format an engine can extract directly. We have not measured its effect on how often a page is cited.

Can franchise businesses work on their AI visibility without rebuilding every page?

Yes. Three actions we would prioritise: adding structured data schema to existing pages, writing unique content for each location, and marking up FAQ content that already exists on the page. None of these require a redesign.

How quickly do changes show up in AI answers?

We do not publish a timeline, because we do not measure one. Telling you a schema change lands in AI answers within a set number of weeks would need per-brand tracking of a change against the answers that follow it, and nothing in Pondral records that today. What the weekly audit does show is your score before and after, dated, with the answer behind each one, so you can see when a change actually moved and by how much.

What's a good AI Visibility Score?

Across 200 brands in Pondral's AI Visibility Index 2026 (a single-run snapshot from June 2026, directional not exact), the average score is 56 out of 100. A score of 50 to 70 is the Established band: consistently cited in AI responses. 70 to 85 is Strong, and 85 to 100 is Dominant, the default reference in your category.

Check your own score

This franchise is a top-ranked brand in its category. Its AI Visibility Score is 38.7. If a top-ranked brand scores 38.7, it is worth checking yours.

You can find out in about a minute. Run a free AI visibility check at pondral.com/demo. The free check runs 2 of the 5 engines we track; paid plans cover all 5: ChatGPT, Claude, Gemini, Perplexity, and Grok.

Continue reading

• The Complete Guide to AI Visibility in 2026
• AEO vs SEO: what changes and what stays the same
• The five-factor rubric, explained slowly
• 5 quick wins to improve your AI visibility this week
• How Pondral scores AI visibility

Published May 16, 2026. Last updated Oct 13, 2026.