,

Where UX Fixes and AI Readiness Overlap

7 mins
Two desktop monitors displaying website layouts, design mockups, and digital interfaces during a web design and UX development process.

Ask most organizations who owns their website experience, and they’ll point to marketing or digital. Ask who owns AI readiness and the answer gets vaguer. Maybe the SEO lead, maybe a working group that formed after someone noticed ChatGPT describing the company in odd ways. Either way, the two efforts usually run on separate tracks, with separate budgets and separate reports.

We recently ran both at once for a global real estate firm, and the amount of overlap surprised us. The UX analysis and the AI readiness assessment were done by different people using different methods, yet they kept landing on the same problems: content tucked away behind clicks, pages with three H1s, images with no alt text, news articles with no author. Each issue frustrated visitors, and each one made it harder for AI engines to understand and cite the site.

That project changed how we think about both kinds of work. A lot of what now gets labelled “AI readiness” is good UX practice that was never quite finished.

Why The Two Keep Running into Each Other

People and AI engines read a website in very different ways, but they stumble over the same things. Someone scanning for a firm’s assets under management won’t dig through collapsed panels to find it, and a crawler parsing the HTML might miss that content entirely.

The qualities AI engines reward are familiar ones. Clear hierarchy, direct answers and credible, named sources have been part of good UX for years, and they line up closely with Google’s E-E-A-T guidelines (Experience, Expertise, Authoritativeness and Trustworthiness). Generative engines pull passages out of pages to answer specific questions, so a page that’s easy for a person to follow is usually easy for a machine to quote.

Five Fixes that Help Both

Content Hidden Behind Clicks

Accordions, tabs and carousels are popular for good reason. They keep long pages tidy and fit a lot into a small space. The trouble is that visitors rarely open them. On the real estate firm’s news hub, a side-scrolling carousel meant most stories were barely seen, and only the first one or two got any real clicks. Investment details and capability descriptions elsewhere on the site were hidden the same way.

AI engines run into a related problem. If content only loads after someone clicks, it may not be in the HTML a crawler reads, and even when it is, hidden text can be treated as secondary. When an engine can’t find a clear answer about what a company does, it goes looking somewhere else, often a directory or a parent company’s annual report.

The fix is mostly editorial. Content that defines the organization belongs on the page, in plain view. Where an accordion still makes sense, its contents should be in the markup from the start.

Heading Structure

This one sounds minor, but it’s one of the most common problems we find. More than half of the main pages in the audit had missing or duplicate H1s, usually because every slide in a homepage carousel had been given its own top-level heading.

Screen reader users move around a page by its headings. AI engines use them to work out which section answers which question. One H1 per page and sensible H2 to H4 nesting fixes both, and it’s about as cheap as accessibility work gets.

Alt Text

When images lack alt text, screen reader users miss the content entirely. In Ontario, that gap also falls under AODA requirements.

It matters for AI too. Engines that index images can’t do much with a photo that has no description, so the firm’s property photography and project imagery, some of its strongest assets, were effectively invisible. Descriptive alt text, backed by image and video sitemaps, brings that content into view for both audiences.

Numbers People Can Actually Read

Animated counters look great in a design review. In practice, they’re easy to scroll past, hard for assistive technology to read, and out of reach for a crawler if the figure lives inside a script or an image. Institutional visitors doing diligence want the number, what it measures and when it was last updated.

When AI engines can’t find those figures on a company’s own site, they borrow them from somewhere else. In the real estate audit, engines regularly went to the parent organization’s reports for basic facts like assets under management, and ended up describing the firm in the parent’s language. Putting headline figures in plain text, with a definition and an as-of date, gives visitors confidence and gives engines something they can cite directly.

People and Bylines

Leadership pages are often among the most engaged on a corporate site. Visitors will spend minutes opening profile after profile, then leave because the profiles offer no LinkedIn links, no related articles and nowhere else to go. The news section often has the opposite problem, with articles that carry no byline at all.

AI engines treat named authorship as a sign of credibility, so an unsigned article carries less weight, and a leader without a profile page has no clear link to the company. Bylines that link to leadership profiles, individual pages for senior leaders, and Person schema tying them together help visitors and engines alike see the expertise behind the brand.

Where the Overlap Stops

Not everything lines up. Plenty of UX work, including calls to action, audience routing, form design and conversion paths, matters a great deal to visitors and very little to an AI engine. A better contact form won’t change how Gemini describes a company.

Some AI visibility and readiness work happens entirely off the page. How a company is described on Wikipedia and LinkedIn, whether it appears in industry rankings, what’s in its llms.txt file, and whether its CDN blocks AI crawlers without anyone noticing all affect visibility without touching the visitor experience.

That’s why we still score the two sides separately in a combined audit, so the places where priorities pull in different directions stay as visible as the overlap.

The Case for Doing Both at Once

The most practical argument is that many of these fixes touch the same templates. Cleaning up headings, surfacing hidden content and adding schema in one pass is far easier than reopening the same components six months later for a separate project.

The data also gets more useful. Heatmaps and journey analysis show what people do on a page. AI visibility tracking, including the searches the engines run behind the scenes before they answer, shows how machines interpret it. Read side by side, each explains things the other can’t.

It also helps with buy-in. When two independent methods flag the same problem, it’s much harder to push it down the backlog.

Getting Started

A combined audit starts with the audience: who visits each main page, and what the business needs that page to do. Traffic matters, but it isn’t the only reason a page deserves attention. Behavioural tracking needs at least a couple of weeks before the patterns are reliable. AI visibility is best measured two ways, with branded prompts that check whether engines describe the company accurately and unbranded ones that check whether it shows up in category conversations at all. The fixes that appear in both reports are the natural place to begin.

See Where Your Site Stands

Our AI Readiness & Visibility Audit can be run on its own or paired with a UX and technical review. Combined, they deliver one clear baseline and one prioritized roadmap.