What did we actually check?
On November 8, 2026, we pulled the raw HTML of 20 Lehigh Valley business homepages and checked the things assistive technology depends on: whether images carry alt text, whether form fields have real labels, whether a keyboard user can skip past the navigation, and whether an accessibility overlay script was present. We counted by hand from the source code — no automated score, no estimates.
We actually examined 32 local sites. Twelve didn't make this report: some couldn't be checked honestly (a site that blocked us, one that never loaded, one built entirely in JavaScript), one was an empty placeholder page, and the rest had findings too thin to report responsibly. Two of the businesses we checked passed everything we tested. Their sites are fine, and there's nothing to sell them. The 20 below are the ones with problems we could verify and reproduce.
One rule before the numbers: we are not naming any business. Every finding here is an anonymized pattern. These are our neighbors, and several of them will fix this — the point of publishing the data is the pattern, not the pile-on.
Problem #1: Images a screen reader can't describe
At least 368 of 630 homepage images — 58% — were undescribed.
Alt text is the description attached to an image in the code. It's what a screen reader speaks aloud to a blind visitor, and it's part of how search engines and AI assistants understand what a business does. Of the 630 images we counted, 66 had no alt attribute at all and 302 had an empty one (alt=""), which tells assistive technology to skip the image entirely. Some sites only let us count one of those two categories, so 368 is a floor, not a ceiling — the real number is higher.
How common was it? On 10 of the 18 countable homepages, the majority of images were undescribed. The worst cases:
- One roofing company's homepage: 93 of 105 images had missing or empty alt text — nearly every project photo on the page.
- One real estate brokerage: 50 of 51 images carried empty alt text.
- One restaurant's online-ordering homepage: all 16 images had empty alt text.
- One law firm: 9 of 10 images undescribed.
- Two homepages — a med spa and a restaurant — came in at 100%: not one described image between them.
These aren't decorative stock photos. They're treatment photos at med spas, project photos at roofers, food photos at restaurants — the exact images that sell the service. To a sighted visitor they look great. To a customer using a screen reader, they're silence.
Problem #2: Forms with fields nobody labeled
Two money forms. 20 fields. Zero usable labels.
A form label is what tells a screen reader "this box is for your phone number." Placeholder text — the gray hint inside the box — does not count; screen readers don't reliably read it as a label, and it disappears the moment someone starts typing.
We found two homepages where the most important form on the site had no real labels at all:
- A real estate brokerage's property-search form: all 14 fields — min price, max price, square footage, keywords — identified by placeholder text only.
- A contractor's quick-quote form: 6 of 6 fields unlabeled, including the name and phone number fields.
Read that second one again. The form that exists to turn a visitor into a lead cannot tell a screen-reader user which box is their name and which is their phone number. That's not a compliance technicality. That's a quote request that never gets sent.
Problem #3: No way to skip the menu
A skip-to-content link is a one-line feature that lets a keyboard user jump past the navigation menu instead of tabbing through every link on every page. It's one of the oldest, cheapest fixes in web accessibility. At least 4 of the 20 homepages didn't have one in their code — and those were only the homepages where we recorded it either way.
The overlay story: paying monthly for a layer that can't fix the code
One finding deserves its own section. One local law firm's website runs a commercial accessibility overlay — the kind of widget businesses pay a monthly subscription for, sold as a quick route to an accessible site.
Does an overlay fix the code underneath it? No. An overlay sits on top of a website like a coat of paint on a wall; it doesn't rebuild the wall. When we checked this firm's actual homepage code, the templates still had no skip-to-content link — a structural gap the overlay doesn't change, because the overlay never edits the templates at all. The business is paying, every month, for the layer — while the code underneath keeps the gaps.
If your site runs an overlay widget, this is the question worth asking your provider: what did you change in my code? If the answer is "nothing, the widget handles it," you're renting a workaround, not owning a fix.
What does this mean for your customers — and your sales?
Strip away the technical language and here's what these three problems do:
- Some of your customers literally cannot complete the journey. They can land on your homepage, but they can't tell what your photos show, can't reliably fill in your quote form, and have to fight your menu on every page. Most don't complain. They leave and buy from the competitor whose site works.
- Machines read your site the same way. Search engines and AI assistants like ChatGPT, Perplexity, and Google's AI answers build their understanding of your business from your code — your image descriptions, your labels, your structure. Sloppy code doesn't just shut out people; it makes you harder for the machines to describe and recommend.
- None of it is visible to you. Every one of these 20 sites looks fine in a browser. That's why the owners don't know — and why "my website looks fine" is not evidence that it works.
How can you check your own website in five minutes?
You don't need to read code to get a first signal. Three quick tests:
- The keyboard test. Open your homepage, unplug (or ignore) your mouse, and press Tab. Can you always see where you are? Can you get past the menu to the actual content without tabbing through dozens of links? Can you complete your own contact or quote form? If you get lost, so do your keyboard-using customers.
- The form test. Open your quote or booking form and click the text next to each field — not the box, the words. If clicking the word doesn't put your cursor in the box, the field probably has no real label connected to it.
- The AI test. Ask ChatGPT or Google: "What's the best [your service] in [your town]?" and "What does [your business] offer?" If you're missing, misdescribed, or a competitor with a cleaner web presence gets the mention — that's your code talking.
If any of those three feel off, your site probably has the same class of problems we found in the other 20.
Get your own numbers
The free scan: we'll run the same hand-check on your homepage and send you the one-page findings report — your real issues, counted, no obligation. Request the free scan →
The Fix Pack — $1,499 flat: we fix the detectable issues in your actual templates — alt text, form labels, ARIA, contrast, focus order, heading structure — across your core pages, with a written before/after report. Founding-client launch: $300 off with code FIRST3 — first 3 clients only, ends November 30, 2026.
The AI Visibility Audit — $99: find out how ChatGPT, Perplexity, Gemini, and Google's AI answers talk about your business — how often you're mentioned, who gets recommended instead, and what to fix first. The full $99 is credited toward the Fix Pack or the AI Visibility Retainer if you upgrade within 14 days of your audit.
Straight talk, because it's our whole brand: the Fix Pack fixes documented, detectable issues in your website's code and reports exactly what changed. It does not certify ADA or WCAG compliance, and it is not legal advice or legal protection. Anyone who sells you a guarantee is selling you something else.
How we got these numbers (and their limits)
- All findings come from the public homepage source code of each business, fetched and counted by hand on November 8, 2026. Counts cover homepages only; a full-site scan will find more.
- Image counts were fully categorized on most sites; where only one failure type was counted, totals here use only what was verified — so 368 is a minimum, not an estimate.
- No business is named, and no finding in this report comes from anything but the business's own public code on that date. Sites change; a business may have fixed an issue since.
- Two of the 32 sites we examined passed every check we ran. When there's nothing to fix, we say so — including to them.