The SEO Expert Behind the Study: What Kyle Roof Learned, and What He Wants to Test Next

For years, businesses have had more choices than ever when it comes to selecting a domain name. Alongside traditional extensions like .com and .net, descriptive domain extensions such as .design, .club, and .courses have given brands new ways to create memorable, relevant online identities.

But while domain choice has evolved, much of the SEO conversation around domain extensions has remained rooted in assumptions rather than data. Do descriptive domain extensions perform differently in search? Does domain length matter? And how much of the conventional wisdom surrounding domain names still holds up today?

Those questions led SEO expert Kyle Roof to partner with us on two controlled studies designed to isolate and measure the impact of domain-related variables on search performance. With more than a decade of experience in SEO and a reputation for rigorous testing, Roof has built his career on challenging assumptions and validating ideas with data. We sat down with him to discuss what inspired the studies, what surprised him most, and what questions he wants to test next.

From Building Websites to Testing Search Rankings

Today, Kyle Roof is known throughout the SEO industry for his testing-first approach to search. Over the past several years, he has run more than 400 controlled SEO experiments designed to better understand how Google's algorithm works. Those findings became the foundation of the SEO tool PageOptimizer Pro, inform the work of his agency, and are shared with marketers through conferences, podcasts, and IMG, the SEO education platform he co-founded.

His reputation for challenging conventional wisdom stems from a simple belief: SEO should be measured, not assumed. One of his most well-known experiments demonstrated this principle by ranking a page filled largely with gibberish text and a handful of target keywords, highlighting just how algorithmic search rankings can be.

But Roof's path into SEO wasn't part of a grand plan.

"I got into building websites as a side project, and marketing came out of that naturally," he explains. "What I found pretty quickly was that I was a lot better at marketing websites than building them."

That realization pulled him deeper into SEO, but it wasn't the marketing side alone that kept his attention.

"Once I got there, the testing side of it really hooked me. I liked that SEO was not just opinion or creativity. It was something you could test. You could change an input, measure the result, and learn what actually moved rankings. That is still the part I love most."

That philosophy – test first, assume nothing – became the foundation for both the gTLD SEO Performance Study and the gTLD & URL Length Ranking Study. 

Why Test Domain Extensions at All?

The inspiration for the studies came as ICANN prepared to open the next application round for new gTLDs. With organizations around the world applying for their own domain extensions, questions about domain name performance are once again top of mind.

"There has been a very persistent assumption in the industry that .com will always win," Roof says. "People tend to treat anything outside the traditional extensions as second-tier by default."

Rather than accepting that belief at face value, he wanted to see what the data would reveal. 

"Once you have descriptive extensions like .art, .design, .florist, or .pet, it makes sense to ask whether those domains might actually be helping in some situations instead of hurting. If an extension reinforces the topic, then it is worth measuring rather than guessing."

For Roof, familiarity and performance are two different conversations.

".com is familiar, yes . . . But that does not mean it is always the strongest strategic choice. If you have a descriptive TLD that lines up tightly with the keyword, the topic, or the brand, that can be a real advantage."

The real question, he says, isn't whether .com remains the default. 

"The right question is, 'What actually performs better when you control the variables?'"

Building a Study That Removes the Noise

The first study set out to answer a simple question: when all other factors are equal, how do descriptive domain extensions perform in Google compared to traditional .com and .net domains? The second study expanded the scope by examining domain length and TLD length as independent variables.

To find out, Roof designed the experiment around a core principle of scientific testing: isolate the variable.

"The goal was to strip away as much noise as possible," he says.

The testing environments used identical WordPress installations, page structures, Search Console integrations, and content. Rankings were tracked using U.S. desktop English search results, and the study incorporated both real-world keywords and artificial keywords to minimize the influence of brand history and other external signals.

"Artificial queries are useful when you want to reduce noise from brand history and other outside signals," Roof explains.

Even in controlled environments, SEO testing isn't always straightforward. According to Roof, one of the biggest challenges was keyword parsing: the process in which a search engine interprets and separates the words and meanings within a query or domain name.

"A domain might look like a clean keyword match to a human, but Google may split it differently,” Roof explains. "We saw that clearly in the singular and plural issues, where the intended keyword match did not line up the way you would expect."

He points to the domain designerb.us, which was intended to target the keyword phrase "designer buses."

"Google split the domain as 'designerb' plus '.us' instead of 'designer' plus 'bus,' so the intended match never registered."

Meanwhile, another test domain behaved exactly as expected: Poptempo.art was split cleanly into ‘poptempo’ and ‘art,’ resulting in the first ranked website for ‘pop tempo art. 

While the studies produced multiple interesting findings, one stood above the rest for Roof. 

"The biggest surprise was how well the descriptive gTLDs performed, especially in partial match setups," Roof says. "That’s where the pattern became hard to ignore. The descriptive extension was not just competitive with the traditional domain. It was the better option."

The domain length research produced another noteworthy trend: both shorter domains, and shorter TLDs, appeared earlier in Google Search results. Taken together, the findings point toward a more nuanced understanding of domain strategy.

"The takeaway is not that one format always wins. It is that there are real advantages outside the .com default, and those advantages show up in more than one way."

Ranking Signals, Trust, and User Behavior

Of course, rankings are only part of the equation. Many brand owners still wonder whether consumer trust and click-through behavior outweigh any search advantage a descriptive domain might provide. Roof believes the discussion is often framed incorrectly. Rather than asking whether trust outweighs ranking signals – or vice versa – brand owners should focus on choosing a domain that is strategic.

"A strong descriptive TLD can improve clarity, memorability, and immediate topical relevance. A domain has to work both for search engines and for people. In many cases, a non-.com can do both very well."

When it comes to choosing a domain, Roof believes company size matters: small businesses often benefit most from flexibility. Descriptive domains can still play an important role in marketing campaigns, product launches, and supporting brand assets.

"If they can secure a domain that is short, memorable, and tightly aligned with what they do, a descriptive TLD can be a smart move. It may actually be the better option than forcing a .com that is awkward, bloated, or less relevant."

What Kyle Would Choose Today

If Roof were launching a brand from scratch today, domain choice would remain a meaningful part of the strategy. His checklist is straightforward.

"I would care about whether the domain is easy to remember, easy to parse, topically aligned, and not unnecessarily long."

And if a descriptive extension helps achieve those goals?

"I would absolutely use it."

After all, he already does. Roof and his team chose IMG.courses when launching their SEO education platform in 2023, and the brand has since grown into a thriving destination for SEO training and community. The choice was pragmatic rather than ideological.

"We use IMG.courses ourselves, so I am clearly comfortable with that choice when it fits the brand. I would not force a .com just because people assume it is safer. In some cases, the non-.com is the cleaner, more strategic option."

The SEO Advice Roof Thinks the Industry Should Stop Repeating

We asked Roof what SEO belief he thinks is most overdue for retirement, and his answer extends well beyond domain names.

"That one-size-fits-all SEO rules are good enough. Usually they are not."

He points to several examples: '.com always wins.' 'Shorter URLs rank better.' 'Exact match domains do not work.' The problem isn't that these statements are always wrong, it's that they're often repeated without context.

"The better approach is to test the claim, look at the conditions, and figure out when it holds up and when it does not. Test – not assume – and focus on what actually moves rankings rather than recycle best practices."

As AI reshapes search experiences, many marketers are wondering whether domain names will matter less in the future. Roof believes their role may evolve, but not disappear. Branding, memorability, clarity, and topical relevance will continue to matter, regardless of how search interfaces change.

"I do not think AI-driven search suddenly makes domain strategy irrelevant. I think it just makes it one part of a larger system."

And in that system, clear and relevant domains still provide value.

Stop Treating Non-.com Domains Like a Compromise

Following the domain studies, Roof already has his eye on another widely debated topic: content management systems.

"There is a long-standing assumption that certain platforms cannot rank, or that Google quietly favors one platform over another."

As with domain extensions, he sees a lack of definitive evidence: "If people are confidently repeating it and the data behind it is thin, that usually means there is a real test there."

If there is one lesson Roof hopes marketers take away from the studies, it’s for marketers and brand owners to stop treating non-.com domains like a compromise. The research suggests that descriptive gTLDs can compete, win, and offer strategic advantages over traditional choices when they align closely with a brand and its topic.

"That does not mean .com is bad," Roof says. "It means .com is not automatically best."

"The right domain choice depends on relevance, clarity, brevity, and how well the whole thing works together."