"No ARIA Landmarks Found" — What That Costs You

Audit findings come in two flavors. Some say “you did this wrong.” Those sting, but at least they point at something. The other kind says “you didn’t do this at all,” and these are easy to shrug off, because absence doesn’t feel like a defect. “No ARIA landmarks found” is the second kind, and I want to spend this essay pricing it out, because I think it’s one of the most underpriced lines on any audit report.

Start with what the finding means mechanically. Landmarks are the labeled regions of a page — the markup that tells software “this part is the navigation, this part is the main content, this part is the search, this is the footer.” They come free with semantic HTML elements like nav and main, or can be added with ARIA role attributes. When a scan reports none found, it means your page presents itself to software as one undifferentiated mass. Everything is there; nothing is labeled.

Now consider what that costs, and for whom. A blind visitor browsing with a screen reader doesn’t take in a page all at once the way a sighted person does. The screen reader moves through content in order, reading as it goes. Landmarks are the escape from that linearity: a keystroke lists the regions, another jumps straight to the one you want. It is precisely the move a sighted visitor makes when their eyes flick to the top-right corner for the login link. With landmarks, finding the main content is one gesture. Without them, it’s wading — through the logo, the navigation links read one by one, the promotional banner, the cookie notice — before reaching the thing the visitor actually came for.

Here’s the part that makes this expensive rather than merely unfortunate: the toll repeats. It isn’t paid once, like a confusing signup flow you struggle through and then forget. It’s paid on every page, on every visit, forever. Navigate from your homepage to a product page? The preamble plays again. Come back tomorrow? Again. A sighted user skims past your header in a quarter of a second without noticing they did it. A screen reader user without landmarks pays a toll of a minute or so, in full, every single time. Multiply a small tax by every interaction and you get a large number. That’s the price. And notice who pays it: not your casual visitors, who leave at the first inconvenience, but your most persistent ones — the people trying hardest to be your customers. A site without landmarks charges its most motivated users the highest toll.

I keep framing this in economic terms deliberately, because the compliance framing undersells it. Yes, landmark structure figures into accessibility standards, and yes, ignoring accessibility carries legal exposure in many places. But a business that adds landmarks to avoid trouble has missed the actual transaction. Screen reader users buy software, book flights, order groceries, and subscribe to services, and they talk to each other about which sites respect their time. A site where everything takes ninety seconds longer than it should is a site people abandon — not in protest, just in the ordinary way anyone abandons a slow checkout line when there’s a faster one across the street. The revenue doesn’t disappear. It goes to whoever labeled their regions.

There’s also a quieter cost, invisible today and growing. Machines increasingly read pages on people’s behalf — search engines have for decades, and now AI agents summarize, compare, and act on web content for users who may never load your page themselves. Every one of these readers faces the same problem a screen reader does: given this wall of markup, what matters? A page with a clearly marked main region answers the question. A page without one forces the machine to guess, and machines guessing about your content means your navigation labels and cookie-banner text getting weighted alongside your actual pitch. Landmarks were designed for assistive technology, but they’ve become how you tell every non-human reader where the substance is.

Against all this, weigh the remedy, because the ratio is the whole argument. Fixing “no landmarks found” doesn’t require a redesign, new tooling, or a consultant. It requires wrapping your primary content in a main element, your navigation in nav, your top banner in header, your bottom matter in footer. On a typical site this is an afternoon, most of it spent finding the right divs to rename. Nothing visible changes; these elements don’t alter styling. When GazeSite grades this finding, the prescribed remedy is nearly always this short — which is exactly why it frustrates me to see the finding persist on site after site. The cost of the problem compounds daily. The cost of the fix is a sandwich-length work session, once.

Absent things are hard to care about. Nobody gets a bug report saying “your regions have no names,” and no screenshot ever looks wrong. But some of the most expensive problems a website can have are the ones that produce no error, just a steady, silent leak of the people trying hardest to use it. This is one. Label your regions. It’s an afternoon. The toll booth comes down for good.