The content is there. The labels are missing.
The basics work: every page answers 200 over HTTPS, http:// and www. both redirect to one address, a missing page returns a real 404, and each program page has its own title and description. What is missing is everything a search engine uses to find, group and trust those pages: a list of them, one address per page, a language label that matches the language, and a machine-readable statement of who the school is and where its campuses are.
The school's own keyword list says it: "medical assistant en español", "phlebotomy en español", "cna en español", "bilingual medical programs". In our searches for Spanish-language medical assistant training in Houston, the results were English-only schools and job ads; no result showed a full program taught in Spanish. That is the opening, and Finlay already has the pages for it.
But every one of those Spanish pages tells Google it is English, carries an English description (almost all an English title too), and is not paired with its English version. Label the Spanish pages as Spanish, list every page in a sitemap, and give each campus its own markup. None of it needs new writing.
Every finding carries a confidence label.
They separate what can be acted on today from what needs an instrument nobody has connected yet. Each finding also shows what the site actually served, so any claim here can be checked without taking our word for it.
Three gaps in what search is told.
All three are settings and page-head code in the site builder, not new content. Together they decide whether Google can list the pages, whether Spanish speakers see the Spanish version, and whether the school is recognized as one organization with five campuses.
The site gives search engines no list of its pages and no single address for any of them
/sitemap.xml answers 200 but contains zero URLs: an empty <urlset>. /robots.txt answers 200 with zero bytes, so it names no sitemap. /llms.txt is also empty.
We found 28 pages only by following links from the homepage. Search engines have to do the same, and pages linked only from deep menus (the Spanish location pages, the Orlando page in Spanish) are the last to be found.
None of the 28 pages has a canonical tag. The homepage is served identically at / and /home, and there are two different contact pages (/contact-191386 and /contact-315915-611205). Without a canonical, Google picks which copy to show and splits the signals between them.
/sitemap.xml 200 107 bytes, 0 <url> entries /sitemap_index.xml 404 /robots.txt 200 0 bytes (no Sitemap: line) /llms.txt 200 0 bytes pages found by following links 28 (plus one Cloudflare e-mail page) canonical tag 0 of 28 / and /home same title, same 1,530 words contact pages /contact-191386 · /contact-315915-611205
The Spanish half of the site is labeled, titled and described in English
The site has a Spanish version of the homepage, all five programs, About, Contact and three campuses: 14 Spanish pages. Their visible text is Spanish (for example the H1 "Programa de Técnico en Flebotomía"), but all 14 declare lang="en" or lang="en-us" and carry the English description of their English twin, and 13 of the 14 also carry its English title.
No page has hreflang links, so Google is not told that /language/es/tecnico-flebotomia is the Spanish version of /phlebotomy_technician. A Spanish-speaking student searching in Spanish is shown an English snippet, or the English page.
The Spanish addresses also follow no pattern: /language/es/asistente-medico, but /language/es/pharmacy_technician; the Spanish campus pages are /houston-location, /san_antonio_location and /dallas-fort-worth-location; and /houston-location-199661 is in fact the Orlando CNA page in Spanish.
URL lang <title> /language/es en-us Medical Assistant, Phlebotomy, CNA, Pharmacy & EKG Programs /language/es/tecnico-flebotomia en Phlebotomy Technician Program | Hands-On Blood Draw /language/es/quienes-somos en Who We Are | Finlay Training Institute /houston-location en Medical Assistant, Phlebotomy & EKG Training in Houston /houston-location-199661 en Entrenamiento de CNA ... en Orlando (H1: "Orlando Campus") Spanish pages with English title + description 13 of 14 hreflang on any page 0
Search cannot tell which "Finlay" this is, or where its campuses are
No page carries structured data. Nothing tells Google or an AI assistant that Finlay Training Institute is a school, that it has campuses at five street addresses with their own phone numbers, or that it offers five certificate programs.
The name is shared. A web search for "Finlay Training Institute" on 2026-11-09 returned other schools — a Finlay medical training institute and a Finlay Institute of Nursing in Tampa, a training company in Australia — and no page from finlayti.com. The menu also links "New York – New Jersey" to a different domain, fmti.edu.
Two of the four schools in our benchmark state all of this in markup: UEI (EducationalOrganization, EducationalOccupationalProgram, contact point) and Fortis (CollegeOrUniversity with address, geo coordinates and opening hours).
JSON-LD on any of 28 pages none web search "Finlay Training Institute" 0 results from finlayti.com (other Finlay schools in FL and AU instead) campuses on /contact-191386 Houston 77079 · San Antonio 78229 · Grand Prairie 75050 West New York NJ 07093 · Winter Park FL 32792 New York - New Jersey menu link https://fmti.edu/en/ search-console verification meta none
Three that cost students before they call.
On a phone, the pages take seven to nine seconds to show their main content
Lighthouse (mobile, lab) scores performance 59–81. The largest paint lands at 6.9 s on the homepage, 9.2 s on the Medical Assistant page, 7.6 s on Houston and 5.1 s on Contact; Google's "good" threshold is 2.5 s.
The homepage downloads 7.8 MB. An embedded YouTube video starts pulling video data (about 1.7 MB) as soon as the page opens, and the program photos are full-size JPGs of 600–880 KB each shown in small frames; Lighthouse estimates 3.6 MB saved by sizing them and 1.4 MB more by modern formats.
On the Medical Assistant page the largest image is marked loading="lazy", which delays the very image the visitor waits for, and the layout shifts while it loads (CLS 0.18; good is under 0.1). The server itself is fast: 50–160 ms.
page perf LCP CLS weight home 75 6.9 s 0.051 7,847 KiB medical assistant 59 9.2 s 0.182 3,603 KiB houston 68 7.6 s 0.025 4,609 KiB contact 81 5.1 s 0.024 1,256 KiB largest files, home video stream 1,780 KB · photos 878 / 750 / 705 / 622 KB LCP image, MA page <img ... loading="lazy"> benchmark best CCI Houston LCP 2.4 s, 666 KiB
Titles and headings that say the wrong thing
The Privacy Policy and Terms and Conditions pages are both titled "Contact Us | Finlay Training Institute" and carry the contact page's description. Search results for the school's legal pages will read "Contact Us".
The Orlando page title promises "Medical Assistant, Phlebotomy, EKG, Pharmacy & CNA Training in Orlando" and repeats itself ("| Orlando Campus (Certified Nurse Aide) Training in Orlando | Finlay Training Institute"), while the contact page lists CNA only for Orlando and the Spanish Orlando page marks Medical Assistant as "Próximamente" (coming soon). A title should not promise programs a campus does not offer yet.
Campus pages have five or six H1 headings each, some of them empty; About has four. One clear H1 per page tells search engines and screen readers what the page is.
The e-mail address is hidden behind Cloudflare's e-mail protection, so crawlers and AI assistants read [email protected] instead of an address.
/privacy-policy/ Contact Us | Finlay Training Institute /terms-and-conditions Contact Us | Finlay Training Institute /orlando Medical Assistant, Phlebotomy, EKG, Pharmacy & CNA Training in Orlando | Finlay | Orlando Campus (Certified Nurse Aide) ... /contact-191386 Orlando: CNA - Certified Nurse Aide H1 count /houston 6 · /orlando 6 · /dallas-fort-worth 5 · /san-antonio 5 · /about 4 e-mail in served HTML [email protected] (Cloudflare obfuscation)
Nobody can tell which page or search a new student came from
The served HTML has no Google Analytics or Google Tag Manager. It loads Cloudflare's visitor counter and a tracking-pixel script. There is no Search Console or Bing verification tag (a DNS record would not show here).
Appointments go through nine different booking calendars on the site builder's booking widget. That can work well, but only if each booking keeps the page and the source it came from.
What we cannot see: which searches show the site, which pages bring students, and how many booked. That needs Search Console and analytics, and the school's booking records.
Google Analytics / Tag Manager not found Cloudflare Web Analytics beacon present site-builder pixel present google-site-verification meta not found booking calendars linked 9 (api.leadconnectorhq.com/widget/booking...)
Seven that ship in an afternoon.
| Win | Detail | Confidence |
|---|---|---|
| Publish the sitemap and robots.txt | Both exist and are empty (Issue 01); the site builder generates them from its page list | Confirmed |
| Retitle the legal pages | Privacy and Terms are titled "Contact Us" | Confirmed |
| One homepage address | /home serves the same page as /: canonical or redirect | Confirmed |
| Clean the homepage description | It contains a line break and repeats the program list from the title; 140–155 characters, one sentence about who it is for and where | Confirmed |
| Drop the meta keywords tag | A 24-line keywords tag on every page; Google ignores it and it tells competitors the target list | Confirmed |
| Fix the Spanish menu on the phlebotomy page | /language/es/tecnico-flebotomia shows the English menu while the other Spanish pages show the Spanish one | Confirmed |
| Publish an llms.txt | Today it is an empty file. A short summary plus the page list, generated from the same list as the sitemap. A secondary experiment, not a ranking factor | Confirmed |
One address, mostly. http:// and www. both answer 301 to https://finlayti.com/, and a missing page answers a real 404. The server answers in 50–160 ms behind Cloudflare.
Program pages that answer the question. Each program page has its own title and description, a single H1, more than 1,100 words, the certification it prepares for and the length of the program. Every page shows a working phone number and a booking button. That is exactly what a student compares; it only needs to be labeled for search.
A good automated score, with the main buttons hard to read.
Schools that serve the public are expected to make their sites usable by people with disabilities, and courts and the U.S. Department of Justice point to WCAG 2.1 AA as the standard. This is context, not a legal assessment. Measured on 2026-11-09 on the English and Spanish homepages, the Medical Assistant page, Houston and Contact with axe-core (WCAG 2.0/2.1/2.2 A and AA rules) and Lighthouse, plus a keyboard-only walk and reflow at 320 px.
The buttons that start an enrollment are below the contrast minimum
The main call to action, "Set an Appointment" (Spanish: "Ver Cursos y Horarios"), is white text on a light cyan button: 2.24:1. WCAG AA asks for 4.5:1 for text of this size.
On the Contact page, the four "Send Us a Message" buttons are orange text on white: 2.8:1. On the Spanish homepage the red "(Próximamente)" labels measure 4.03:1.
These are the buttons a prospective student needs most, and people with low vision or on a phone in sunlight are the first to miss them.
element text background ratio needed "Set an Appointment" #ffffff #00bde0 2.24 4.5 "Ver Cursos y Horarios" #ffffff #00bde0 2.24 4.5 "Send Us a Message" (x4) #f47715 #ffffff 2.80 4.5 "(Próximamente)" #e93d3d #ffffff 4.03 4.5
Headings do not describe the page structure
Campus pages have five or six H1s, including empty ones; the homepage jumps from H2 sections to H4 and H5 (program names, staff names, certification bodies). Screen-reader users navigate by headings, and empty or skipped levels make the page hard to scan.
Most content sits outside any landmark region (27 to 32 elements per page), so "skip to main content" style navigation has nothing to land on.
H1 per page /houston 6 · /orlando 6 · /dallas-fort-worth 5 · /about 4 empty H1s present on campus pages heading-order home 4 · Spanish home 4 · medical assistant 2 · houston 1 content outside home 27 · MA 30 · houston 32 · contact 15 elements landmarks
Photos that carry meaning are marked as decoration
On the homepage, 39 of 58 images have an empty alt="", which tells screen readers to skip them. That is right for decorative shapes, but it includes the program photos and the largest image on the Medical Assistant page.
Automated tools accept an empty alt as valid, so this does not lower the Lighthouse score. A person has to decide which images say something.
images 58
alt="" (decorative) 39 includes program photos
no alt attribute 1
LCP image, MA page alt=""
| Barrier | Pages | WCAG | Severity |
|---|---|---|---|
| Button text below 4.5:1 contrast | home 1 · ES home 3 · MA 1 · contact 5 | 1.4.3 | Serious |
| Several H1s per page, some empty; heading levels skip | campus pages · home · MA | 1.3.1 · 2.4.6 | Moderate |
| Content outside landmark regions | all pages checked | 1.3.1 | Moderate |
| Spanish pages declared as English (lang="en") | 14 Spanish pages | 3.1.1 | Serious |
| Meaningful photos with empty alt | home 39 of 58 images | 1.1.1 | Moderate |
| Touch targets under 24 px tall (footer links, menu toggle) | home 9 · contact 22 | 2.5.8 (2.2) | Minor |
Lighthouse accessibility scores 95–98 on the four pages checked. Pages reflow at 320 px with no sideways scrolling, every keyboard stop we reached showed a visible focus outline (except the embedded video frame), and the phone number is a real tel: link on every page.
Limits. On the homepage the keyboard walk stopped at the embedded YouTube player after 16 stops; what follows it was not walked. No screen-reader session was run, and no form or booking was submitted, so error messages were not seen. A person should test with VoiceOver or NVDA before any compliance claim is made.
Four Houston schools, one instrument.
Each page was read with the same script and the same Lighthouse run used on finlayti.com, on 2026-11-09: three schools that appear in searches for medical assistant and phlebotomy training in Houston, and one phlebotomy-only school. For the multi-campus schools we measured their Houston campus page. Measured markup, not opinion.
| Signal | Finlay | CCI Training | UEI College | Fortis College | AccuDraw |
|---|---|---|---|---|---|
| Page measured | homepage | Houston page | Houston page | Houston North | homepage |
| Sitemap lists the pages | 0 URLs | Yes | Yes | 1,007 URLs | Yes |
| robots.txt names the sitemap | Empty file | Yes | Yes | Yes | Yes |
| Canonical tag | None | Yes | Yes | Yes | Yes |
| School / campus markup | None | Breadcrumbs only | School + programs | College + address + hours | Organization |
| llms.txt | Empty | Yes | Yes | 404 | Yes |
| Mobile performance (lab) | 75 | 98 | 68 | 41 | 47 |
| Largest paint, mobile | 6.9 s | 2.4 s | 3.5 s | 22.2 s | 8.1 s |
| Page weight | 7.8 MB | 0.7 MB | 0.4 MB | 12.0 MB | 4.1 MB |
| Teaches in Spanish | Yes | Not stated | Not stated | Not stated | Not stated |
Domain authority and keyword rankings. We hold no backlink-tool license and none has been validated here. Scraped search positions vary by location, history, device and minute; position data comes from Search Console once it is connected.
"Teaches in Spanish" means the measured page says so; a school may offer Spanish classes without saying it on that page.
What the benchmark actually says
- Finlay is the only one without the basics. Every other school publishes a sitemap, names it in robots.txt and sets canonical tags. These are settings, not content.
- Markup is where the larger schools pull ahead. UEI describes each program in markup; Fortis gives each campus its address and hours. Finlay has five campuses and says it in none.
- Speed is a weakness across the field. Only CCI and UEI paint in under four seconds on a phone. Getting there is the fastest way for Finlay to be better than most of its competitors on something students notice.
- Spanish is Finlay's alone. None of the pages measured says it teaches in Spanish. Finlay does, and buries it under English labels.
Grouped by what the student is looking for.
No keyword tool and no Search Console yet, so volume and difficulty are Not measured. These come from the programs, campuses and languages the site describes, and from the school's own keyword list. Once Search Console is connected, this section is replaced with real query data.
Brand
The site should own its own name before anything else (Issue 03).
- finlay training institute
- finlay training institute houston
- finlayti
- finlay school medical assistant
- finlay training institute reviews
Programs by campus
The money searches. Each needs a page that names both the program and the city.
- medical assistant school houston
- phlebotomy classes houston
- cna classes dallas
- ekg technician training san antonio
- pharmacy technician program houston
Spanish-language programs
The wedge. Only works once Issue 02 is fixed.
- curso de asistente médico en español houston
- clases de flebotomía en español
- cna en español dallas
- técnico de farmacia en español
- escuela de enfermería en español texas
What students ask · AEO
Questions AI assistants are asked; the program pages already answer most of them.
- how long is phlebotomy training in texas
- how much does a medical assistant program cost
- do i need a certification to be a medical assistant in texas
- nha vs amca certification
- what does an ekg technician do
Orlando and New Jersey
Only once those campuses' programs are open and the fmti.edu relationship is clear. Hypothesis, not a measurement.
- cna classes orlando
- cna en español orlando
- medical assistant school west new york nj
- patient care technician program nj
Measurement before the work that needs measuring.
Instrument first
- Verify finlayti.com in Google Search Console and Bing Webmaster Tools and give CleverWave read access. It is the only source of real searches, clicks and indexed pages.
- Google Analytics 4 with bookings and calls as conversions
- Confirm who edits the site builder and how changes are approved
- Confirm the campus facts to publish: addresses, phones, hours, programs open per campus, and the relationship with fmti.edu
Give search a map
- Publish the sitemap with every English and Spanish page; robots.txt with the Sitemap line
- Self canonical on every page; one homepage, one contact page
- Retitle Privacy and Terms; fix the Orlando title
- Submit the sitemap in both consoles; IndexNow if the builder offers it
Make the Spanish half Spanish
- lang="es", Spanish titles and descriptions on the 14 Spanish pages
- hreflang pairs on every English/Spanish twin
- One Spanish URL pattern, with 301s from the old addresses
- School, campus and program markup (Issue 03)
- Button contrast, one H1 per page, image descriptions (ADA section)
Speed, then depth
- Click-to-play video, resized WebP photos, no lazy-loaded first image
- One page per program per campus where the campus offers it, in both languages
- A short guide per student question (Tier 04), in English and Spanish
- Replace the keyword tiers with real Search Console queries
- Core Web Vitals verified against field data
The row that answers the question is blocked.
Everything we can measure from outside is scripted and re-runnable, and every number in this report reproduces from the evidence saved with it.
# Crawl of every page linked from the homepage, plus sitemap, robots, llms.txt, 404 and redirects artifacts/clients/finlayti/raw/crawl-2026-11-09.json # Lighthouse 12.8.2 mobile: home, medical assistant, Houston, contact; benchmark pages artifacts/clients/finlayti/raw/lh/ # axe-core, keyboard walk, 390 px and 320 px checks artifacts/clients/finlayti/raw/a11y-2026-11-09.json # Same-instrument benchmark artifacts/clients/finlayti/raw/benchmark-2026-11-09.json
| Metric | Today | Instrument | Available? |
|---|---|---|---|
| URLs in the sitemap | 0 | curl | Now |
| Pages with a canonical tag | 0 / 28 | crawl | Now |
| Spanish pages with Spanish lang, title and description | 0 / 14 | crawl | Now |
| Pages with structured data | 0 / 28 | crawl | Now |
| Largest paint, mobile (lab) | 5.1–9.2 s | Lighthouse | Now |
| Main button contrast | 2.24:1 | axe-core | Now |
| Indexed pages, searches, clicks, positions | — | Search Console / Bing | Blocked |
| Bookings and calls by page and source | — | analytics + booking records | Not measured |
| Core Web Vitals (field) | — | CrUX | Not measured |
Only Search Console and the booking records can say whether more students find and book. Everything above it proves the site changed. That is why Phase 00 comes first.
What this document does not know.
- Coverage. The 28 pages reachable by links from the homepage. With an empty sitemap, pages linked from nowhere could exist and would not be in this count. The student portal and the booking widgets were not opened.
- Served, not rendered, for the crawl. Lighthouse and axe-core rendered five pages in a mobile browser.
- Search visibility was checked with web searches, not with Google or Bing data. "No results" is a reading from one tool on one day.
- No rank, no volume, no authority scores anywhere in this document, by design.
- No legal or accreditation assessment. We record what the pages say ("licensed and approved"), not whether it is current. Compliance with the ADA is not assessed; the accessibility section reports what was measured.
- Identities. We name no staff member, founder or student from the site.
No form or booking was submitted, no account was created, and nothing on the site was changed. Requests were spaced more than a second apart.
Provenance. Measured 2026-11-09. Read-only throughout.
Stop diagnosing,
start building.
CleverWave builds deep SEO audits for organizations that need to be found by the people they serve. We measure every flaw against the served bytes, prioritize by impact, and deliver a phased plan calibrated to your team. Phase 00 takes an hour and starts the clock on everything else.
Book the conversation