By Maggie Morrissey, PR director at specialist B2B marketing agency, Lesniak Swann.

Specification in the fenestration sector has always been heavily driven by performance and compliance risk. What’s changing now is how that process begins. According to RIBA, 59% of architects are using AI tools in their work, whether that’s to sense-check options, compare systems or build an initial shortlist before speaking to suppliers.

This change in behaviour is drastically changing the way online search works. Try this quick search. Type “Who is the world’s leading supplier of aluminium windows?” into Google. The result at the top of the page is likely to be an AI-generated summary, naming two or three companies. Reynaers Aluminium will almost certainly be one of them. Increasingly, for many users, that’s where the search journey ends, without ever clicking on a link.

This is fundamentally different to the way traditional search has worked for years. That’s because AI summaries don’t rank pages based on links and keywords. They bring together information that is repeatedly cited across sources that appear authoritative. Brands that show up consistently in those environments are more likely to be included.

This is also where a lot of the current marketing advice misses the point. There is a growing focus on optimising content for AI, whether that’s through the use of structured data, FAQs or Generative Engine Optimisation (GEO) tactics. These kinds of tactics are a hangover from the way brands have been using traditional search engine optimisation (SEO) over the last 30 years.

However, that approach assumes visibility is driven by formatting, or by optimising the copy on your website for individual search terms. It isn’t. Inclusion in AI summaries comes from being consistently present in credible, relevant conversations across a broad range of sources.

For fenestration businesses, this means the value of trade press matters now more than ever before. Non-salesy articles that explain system performance, respond to regulatory change or address common specification challenges leave a lasting impact. They become part of that wider set of authoritative sources that AI systems draw upon when generating answers.

We have seen this in practice through our work with Reynaers. Rather than relying on one-off product announcements, the focus has been on sustained engagement with the market over many years.

That has included regular technical commentary on Part L of the building regulations, the thermal performance of buildings, sustainability targets and system design. Over time, that has built a clear association between Reynaers and the issues that specifiers, fabricators and installers face day to day.

As Samantha Hill, head of marketing at Reynaers Aluminium UK, explains:

“In the glazing industry, we often focus heavily on technical performance indicators – U-values, insulation ratings, and sustainability metrics – but without clear explanation and context, these figures can be confusing or even meaningless.

“When they aren’t communicated consistently, they’re far more likely to be misinterpreted, especially when condensed into short, simplified answers by AI tools.”

Brands that have been working in this way all along, rather than optimising their traditional search engine rankings for individual search terms, will have greater visibility in the new era of AI search, where information from multiple authoritative places is being pulled into a single summary answer.

It’s important to caveat that there is still a role for technical optimisation. Clear structure and accessible content will always help. But those elements are the baseline. They don’t establish authority on their own. That comes from consistent contribution to credible, specialist conversations over time.

If specifiers are increasingly turning to AI-generated summaries, then the sources behind those summaries carry more weight. The question is no longer whether a specifier will click through to your website. It’s whether your brand is part of the answer they see before that process even begins.