Web Accessibility Habits Are Paying Off in the Race for AI Citations

Buyers increasingly put a full question to an AI assistant and get back a short list of recommended names, and whether a brand makes that list depends on whether a model can parse its pages cleanly. As more people hand purchases to assistants, the work that decides who gets surfaced turns out not to be new. It's the same accessibility work teams have deferred to the end of every website build for twenty years, the structure a page needs so software can read it without seeing it. What used to be a late box to check now decides whether an answer engine can quote you at all.

Jessica Kerrigan is a fractional CMO who builds brand and demand programs for B2B SaaS and tech founders, often as their first senior marketing hire. She spent close to two decades across B2B SaaS and consumer brands, and started as a writer, editing a motorcycle magazine before moving into marketing. At Cecilian Partners she built the marketing function from close to scratch and grew its pipeline contribution past $3 million. She keeps returning to one idea that predates any of the current tooling: whatever the channel, marketing is one human being trying to reach another.

That instinct shapes how she reads the arrival of AI search, which she frames as an old problem wearing new clothes. "Design is important to the human who can see it and interact with it. But when you strip all of that away, what a machine is actually reading is just structure and meaning," says Kerrigan. The systems doing the reading now, she argues, want the same thing a screen reader has always wanted.

One read, two audiences

Her route into this was noticing an overlap most marketers miss. "It was so interesting to make the connection between AI and accessibility, because it comes down to something that has never changed," says Kerrigan. "Making your site readable to the AI machines is all about structure, clarity and meaning, which is the same thing you're doing to make it accessible to any person on the planet." A screen reader, the software that reads a page aloud to a blind visitor, doesn't process the visual layout at all. It moves through the underlying markup: headings, link text, labels, and the order things appear in the code.

An AI crawler indexes a page along the same lines, which is how screen readers navigate content that existed long before answer engines did. "A screen reader and an AI crawler both read your website. Neither one is really seeing it," she says. For Kerrigan that reframes what design is for. It still matters to the person on the page and to the brand, but it's invisible to the systems deciding whether to surface you.

The floor came up

The catch, in her view, is that broad access to AI has made competent content ordinary. "AI has raised the floor, and it's made it harder to stand out when a lot of content now sounds the same," says Kerrigan, describing a web where content generated at scale has flattened the quality baseline. Clean structure gets a page through the door with the machines, but it does nothing to keep a reader once they're there.

That's where the human half of the work earns its keep, and where showing up in answers starts to depend on more than markup. "The thing that's going to distinguish you is knowing your audience better than anyone and speaking to that," she says. "The structure will get you read by the machines, but it's the point of view that's going to resonate with the human being."

Build it once, together

Kerrigan points to a heuristic she's heard repeated in accessibility circles: get a page to accessible, and you're roughly 70 to 75 percent of the way to machine-readable as well. The problem, she says, is sequence. Teams still ship the pretty version first, hand it to an accessibility reviewer second, then ask a content lead to make it stand out third, each pass quietly undoing part of the one before it.

Her fix is to collapse those passes into one. "One of the most important things we can do is consolidate these workflows and have teams working together from the beginning, so everything that matters is baked in from the start," says Kerrigan. What holds that collaboration together, she adds, is a single question everyone keeps in front of them. "The teams I've seen work best are the ones where everyone, from the developer to the designer to the writer, is asking, 'Am I making life easier or harder for my website user?'"

Be the best human

If software now handles the baseline, her advice is to pour effort into the part it can't reach. "When AI can do your proofreading and some of the best-practice content, then it comes down to you to be the best possible human you can be in connecting with other human beings," says Kerrigan. The durable skill, she says, is the judgment machines lack: "It's your ability to actively listen to people, to glean insights and to synthesize those into something meaningful."

She expects the vocabulary to shift as the payoff stops looking procedural and starts looking commercial. "We're going to start dropping the word compliance," says Kerrigan. "It's not about complying anymore. It is about building a bigger, better served audience that converts." The structural work and the human work, in her telling, are two halves of one job. "The machine needs structure, and the people need your unique voice and your unique point of view," she says.