← Part of the full guide: Why AI Doesn’t Recommend Your Brand
TL;DR: Your site builder sets a hard ceiling on how high your AI visibility score can go. In one of our cases, a Wix SPA with a 6.75% HTML-to-code ratio topped out at 72 of 100 after every available fix had been made. This page explains where the ceiling comes from, how to measure your own, and the criteria for deciding when migrating is worth it and when it is not.
What a ceiling is
Most AI visibility work assumes you control what gets served to a crawler. That assumption often fails. Hosted site builders limit how much of the document head you can edit, whether structured data survives, and how pages render. Past that limit, the score stops moving no matter how much work continues to be billed.
This matters because it is the quietest way a budget leaks. Clients pay month after month chasing points the platform will never allow, with nobody telling them where the finish line actually sits.
Two real cases that hit a ceiling
| Case | Before | After | What capped it |
|---|---|---|---|
| Bodymap TH — Wix SPA | 38 | 72 | 6.75% HTML ratio, putting the ceiling near 72 |
| Coldtubb | 37 | 55 | Technical Performance moved only 4 points, 46 to 50 |
The Coldtubb case is worth reading closely. The dimensions driven by content moved sharply — AI Citation Presence 25 to 65, AI Overview Presence 0 to 40 — while Technical Performance barely shifted because the platform owns it. Judging by the total alone would suggest weak results, when in fact everything controllable was fully executed. Per-dimension figures are on the AIVS methodology page.
Where the ceiling comes from
HTML-to-code ratio
The most direct indicator is what share of the served document is actual readable text rather than markup and script. Sites that wrap their copy in heavy JavaScript scaffolding leave very little real text behind. In the Bodymap case that figure was 6.75%, meaning almost the entire file carried no meaning for a model.
How much of the document head you control
Some platforms restrict which meta tags and structured data you can add, or accept them and then overwrite them with system defaults. When schema does not survive to the served page, Content Understandability cannot reach the score it otherwise would.
Where rendering happens
If text only appears after JavaScript executes, the major AI crawlers never see it, because they do not run JavaScript. Server-rendered or pre-built static pages avoid this entirely.
Whether you can write files at the domain root
robots.txt, sitemaps and files such as llms.txt must sit at the root of the domain. Platforms that do not expose these directly place a hard limit on the Crawlability dimension.
Measure your own ceiling
- Measure the real-text ratio. Fetch a page with
curland compare the character count of readable copy against total file size. Below roughly 10%, assume a ceiling exists. - Test whether schema survives. Add Organization schema, then re-fetch with
curl. If it is not in the response, the platform is not serving it to crawlers. - Check root-level file access. Try editing robots.txt and adding a new file at the domain root. If you cannot, Crawlability has a hard limit.
- Diff what humans see against what bots see. Open the page in a browser and compare it to the
curloutput. Whatever is missing is invisible to these systems.
Run these four before committing to any long-term engagement, because they determine the maximum result the spending can produce.
When migrating is worth it, and when it is not
Worth considering
When the score has plateaued well below competitors and the business depends heavily on search-driven discovery. In the Bodymap case, moving to WordPress or Next.js would unlock the 82–88 range against a current ceiling of 72. That 10 to 16 point difference has to be weighed against migration cost and the risk of breaking what already works.
Not yet
When everything achievable on the current platform has not been done. Many sites that appear capped have not yet rewritten the homepage to name the product clearly, added schema, or reconciled their business details. That work is possible on almost any platform and usually moves the number more than expected.
Migration will not help
When the blocker sits off-site. If a business cannot be corroborated from external sources, migrating raises the structural score while the business still goes unrecommended. We ran into exactly this ourselves — full control over our own site, still never named — as documented in our self-test.
The question to ask a provider first
The single most budget-saving question is: what is our ceiling on the platform we are on today. A good answer gives an approximate figure and names the constraint producing it. An answer worth treating carefully is that there is no ceiling, because every hosted platform imposes some limit.
The full screening set is at seven questions before signing.
CiteLogics — the ceiling is disclosed before you spend
The AI Visibility Audit states the ceiling imposed by your current platform in the report itself, along with an estimate of what migrating would unlock, so the decision happens before a long-term budget is committed.
- AI Visibility Audit — full AIVS Score report plus a 30-minute call. THB 2,000, delivered in 24 hours, 7-day money back.
- Big Win (Audit + WebFix) — audit plus full implementation across crawlability, structure and schema. THB 20,000.
Get your free AI visibility scan, or contact sales@citelogics.com.
Frequently asked questions
Which platforms have ceilings?
It depends on the individual build more than the platform name. Two sites on the same tool can have different ceilings depending on theme and plugins. Measure the real-text ratio in the served document rather than inferring from the vendor’s name.
What HTML ratio counts as a problem?
In the case we measured, 6.75% produced a ceiling around 72 points. If your real-text share falls below roughly 10%, assume a ceiling exists and assess it before planning a long-term budget.
Does migrating raise the score immediately?
No. Migration creates the possibility of work that was previously impossible, but that work still has to be done, and pages must be recrawled afterwards. Plan the migration and the structural work as one project rather than two.
What can I still do once I have hit the ceiling?
Off-site work is unaffected by platform limits — business listings, reviews and third-party mentions all remain fully available. In the Coldtubb case the citation dimension moved from 25 to 65 while the technical side stayed capped.
When should a provider raise the ceiling issue?
In the first audit report, before proposing any long-term engagement. Not raising it leaves the client paying continuously to chase points the platform will not permit, which is where budgets leak without anyone noticing.
Written by
Nipun Kasevayuth
Founder & CTO, CiteLogics
An engineer and AI enthusiast who reverse-engineers how AI models decide which brands to cite. He proved the method on his own websites first — then delivered the same results for brands like Coldtubb. He and the CiteLogics team have scanned 1,000+ websites for AI visibility.
Connect on LinkedIn →
