Somewhere today, a founder is staring at a domain reseller's price tag, trying to decide if $3,000 for "bestplumbingchicago.com" beats the brand name they actually like. The pitch hasn't changed in years. A keyword in the domain helps you rank. Lately it comes with a second promise too, that it'll help ChatGPT and Google's AI answers cite you as well.
Somebody finally ran the numbers instead of repeating the folklore.
Google already penalized this once
This isn't a new argument. In September 2012, Google shipped an algorithm update built to strip ranking power from low-quality sites leaning on a keyword-stuffed domain instead of real content. It hit thin affiliate sites and content farms hard. "EMD" became shorthand for a dead tactic across most of SEO for the next decade.
Google's own current documentation never fully closed the door, though. Its public ranking systems guide still lists an "exact match domain system" among the signals that factor into relevance, described as one of many rather than a shortcut. That's the exact gap one practitioner's test was built to measure directly, fourteen years after the original penalty, with AI engines now added to the question.
A $224 test beats a hunch
Krešimir Ćorluka's study for Ovisy, the SEO and AEO agency he runs, came out on November 2, 2026. He pulled 7,374 commercial keywords across the US and UK in 10 industries, reading the top 20 results for each one. That's 106,130 results in total.
He added 6,267 local keywords like "plumber tulsa" across 40 cities, pulling each city's results from inside that city rather than from one national vantage point. Then he checked 3,048 domains registered in 2024 and 2025 again, 16 to 29 months later, to see which ones actually ranked.
On top of that, he ran about 60,000 questions through three AI models, GPT-4.1 mini, Gemini Flash and Claude Haiku, to see which domains they'd cite or recommend. All told, the project cost him $224 in API credits.
His own summary, after all that, was blunt. A keyword in your domain isn't going to do much for you.
What an exact-match domain buys you on Google
Some lift exists, just a small one. Weak exact-match domains, the kind with little authority yet, ranked about 0.9 positions higher on average than comparable brand domains at the same Domain Rating. Their odds of reaching the top three came in roughly 10 percentage points better.
That edge mostly belongs to older sites. Once Ćorluka isolated domains younger than five years, the advantage stopped being statistically reliable. A brand-new exact-match domain isn't inheriting any of that lift yet, no matter how perfectly it matches the keyword.
Local search is the one place it almost shows up
"Tulsaplumber.com" style domains behaved a little differently from the commercial set. At matched authority, they weren't any more likely to crack page one than a plain brand domain. But once a keyword-match local domain did land on page one, it showed up in the top three noticeably more often. Call it about one extra position, on average, once you're already in the running.
Reddit's local SEO crowd didn't fully agree when the study got cross-posted to r/localseo. One agency owner described building two sites for the same type of business two years back, one branded, one a city-plus-service domain. Both carried similar authority, Domain Rating 26 versus 29, and neither had been touched since launch.
By their own account, the city-plus-service domain ranked substantially higher for almost every keyword and showed up more consistently in the map pack, the opposite of a null result. Another commenter argued the domain was beside the point entirely. Stuff the keyword into your registered business name instead, they said, and the map pack rewards that far more than any URL choice ever will.
Treat both reactions as single data points rather than a rule. The commercial test ran across thousands of domains. A two-site comparison or a strong opinion about map pack behavior is real evidence, just much thinner evidence.
26 content formats. Published on your domain. Matched to your brand.
Guides, comparisons, listicles, case studies, and more. RankControl generates content that gets cited by ChatGPT, Perplexity, Claude, Gemini, Grok, Google AI Mode, and Google AI Overviews.
A new domain doesn't get a head start, keyword or not
Among domains registered in 2024 and 2025, only 4.7% of the exact-match ones reached the top 10 for anything at all, 16 to 29 months after registration. Other new domains in the same niches did a touch better, at 5.7%.
Narrow it down to the domain's own target keyword, once that keyword had real monthly search volume, and just 0.7% of new exact-match domains made the top 10 for it.
Zoom out further and one number matters more than the rest. Ninety-four percent of every new domain in the study, keyword-matched or not, ranked for nothing at all in that window.
One AEO forum reader summed up the reaction to that figure well. People keep treating a domain name as a growth hack, they said, when the real bottleneck is almost always distribution and authority.
Building that is the part a flashy domain can't shortcut. Publishing on a schedule and earning real links and mentions over time is still what moves a new site out of that 94%, whatever the URL says.
A keyword in the URL sways AI models a little
Two separate tests here, with two different answers. First, when the three AI models searched the live web for an answer, they cited exact-match domains at roughly the same rate Google ranked them. No extra AI-specific bonus sits on top of the ranking bonus.
Second, Ćorluka ran a more direct test. He gave each model five equally good pieces of writing, each one attached to a different, randomly generated domain name, and asked which one it would recommend.
The exact-match name got picked 59% of the time, against a 20% baseline if the choice were random. A model does seem to carry some bias toward a keyword-shaped name once it's already weighing similar options.
One number is worth remembering before anyone pays a premium for that bias. A keyword placed in the URL path of an ordinary brand domain, on a page the model had already decided to pull up, scored almost as well, at 48%. Most of that bias is available for free, on a domain you already own.
Your competitors are getting cited by AI. You're not.
Every day without citation tracking is a day your competitors pull ahead in ChatGPT, Perplexity, and Claude.
What the test couldn't rule out
Ćorluka listed his own study's limits alongside the results, which is rarer than it should be in this space. The study is observational, so it can't fully separate a domain's keyword from the site's overall quality.
Exact-match domains in his sample were mostly bought by people already running SEO campaigns, so those sites may simply carry a better-optimized site behind a better-matching name. He also checked only the top 20 results and measured each domain once rather than over months.
The AI side ran through raw APIs too, rather than the consumer ChatGPT, Gemini and Claude apps most buyers actually compare against.
None of that erases the pattern his numbers show. It's why "a small, mostly unreliable lift" is a more honest read of this data than a flat yes or a flat no. Nobody should treat a single test, however large, as the final word on a question SEOs have argued about since that 2012 penalty.
The DIY math versus letting someone else run it
If you already own a branded domain and you're tempted to migrate to a keyword-match one chasing this edge, add up what that actually costs first. Mapping and setting up 301 redirects for every URL on a mid-sized site usually runs 10 to 20 hours on its own.
Re-pointing backlinks and updating business listings is its own project on top of that, and watching rankings through the reindex adds still more time, spread over several weeks while Google works out what the new domain is. For most sites, that's more time and more risk than the roughly one-position gain this data supports.
Put that same budget into content that actually answers what buyers search, and into the links and mentions that build real authority instead. RankControl writes that kind of content against the pages already ranking for each keyword and publishes it on your own domain through our content engine.
Our outreach agent finds link prospects for whatever gets published and drafts the emails from your own mailbox. You can do all of this yourself, URL change included, with the same budget. Or let RankControl run it every week while you work on the business itself.
What the domain name is actually worth
A little, mostly for older or local sites, and not enough to pay a premium for on its own. If the exact-match name is also one you'd be happy to put on a business card, buy it and move on. If it isn't, keep the brand name you actually like and spend the difference on content and links.
The real work starts after this one decision. Whatever domain you pick, you still need to know, week after week, whether it's gaining ground in Google and in AI answers, since both keep shifting long after launch day. RankControl's visibility tracking checks that across seven AI engines, so the next domain decision rests on real numbers.
200+ SaaS teams already track their AI citations.
They know exactly when ChatGPT mentions their brand, and when it stops. Do you?
