This site is a record, not a business. BillHartzer.net publishes Bill Hartzer’s professional history and practice. It sells nothing, quotes nothing and takes no engagements — consulting and expert witness inquiries go to Hartzer Consulting.

BillHartzer.net logo mark — the professional record of Bill HartzerBillHartzer.netThe professional record of Bill Hartzer
Search Engine Optimization

Top-Level Domain Bias in Search

Whether your extension affects indexing and rankings, where the real effects are, and why most studies of this prove nothing

Abstract stacked chevron illustration representing Top-Level Domain Bias in Search

Stating the question properly

“Do domain extensions affect rankings” is asked constantly and answered badly, usually because two different questions are being run together.

The first is whether a search engine applies a ranking adjustment based on the string after the final dot — whether .com receives a bonus, or a newer extension receives a discount, as a direct function of what it is. The second is whether sites on particular extensions tend to perform differently for reasons that travel alongside the extension: who registers there, what they build, how users respond, and whether the surrounding population is dominated by abuse.

The first question is about mechanism. The second is about population. They have different answers, and conflating them is why this argument never resolves. It is also why I have spent a fair amount of time on it — I have written about new generic extensions and search several times, including for Search Engine Journal in 2016, and the pattern I keep coming back to is that people looking for a ranking factor are looking in the wrong layer.

What is actually claimed, and by whom

Google has stated publicly and repeatedly that the newer generic top-level domains are treated like any other generic extension, that they carry no inherent ranking advantage or disadvantage, and that keywords in the extension do not function as a ranking signal. That position has been consistent since the new extensions began launching in 2013.

Two documented exceptions sit alongside it. Country code extensions are used as a geographic targeting signal: a .de or a .fr tells a search engine something specific about the intended market, and that is deliberate rather than incidental. And a small number of extensions are treated as generic despite technically being country codes, because they were adopted globally for reasons that had nothing to do with their territory.

Against that, practitioners keep reporting that new extension sites index more slowly, rank more reluctantly, or appear to sit under some kind of ceiling. Those reports are not fabricated. The question is what produces them, and there is a more mundane explanation available than a secret multiplier.

Indexing is where the difference shows

The more interesting layer is not ranking but indexing, and it is the one I examined in an article for CircleID published on June 18, 2020, titled “Top-Level Domain Bias in Search Engine Indexing and Rankings.” The approach there was to remove the confounding factors rather than argue about them: register the same invented, meaningless word across fifteen different extensions, publish comparable content on each, and then watch what the search engine did with them. Identical strings, identical content, no history and no brand on any of them — so any difference in how quickly and how completely they were indexed had to come from somewhere other than the sites themselves. The extensions did not perform alike.

Ranking is a comparison among pages already in the index. Indexing is a prior decision about whether a page is worth storing at all, and that decision is made under resource constraints using whatever signals are available at the time. For a brand new site with no links, no history and no recognizable brand, the available signals are thin. A crawler encountering an unfamiliar host has to make a judgment about how much attention it deserves, and everything it knows about the neighborhood that host sits in is part of that judgment.

Which means the effect people experience as a ranking penalty is more often a discovery and retention problem. Pages take longer to be fetched. Fewer of them are kept. The site appears to underperform, and the extension gets the blame for what is actually a cold start with an unusually thin set of signals. The practical implication is different from the theoretical one: if you launch on an unfamiliar extension, the work that matters most early is the work that establishes the site exists and is real — links from places already crawled frequently, a clean sitemap, an unambiguous entity, and internal linking that gives a crawler a reason to keep going.

Where the effects are genuinely real

Set the algorithm aside and several effects remain, all of them consequential and none of them mysterious.

  • Geographic targeting. A country code extension genuinely helps inside its market and genuinely constrains you outside it. This is documented behavior, not inference.
  • Abuse concentration. Some extensions have unusually high proportions of spam, malware and phishing registrations, largely because they are cheap or were promoted at near-zero cost. That reputation attaches to the extension in the systems that matter most immediately: mail filters, security products and browser warnings. It affects deliverability before it affects anything else.
  • User behavior in the results. A person scanning results makes a trust judgment in a fraction of a second, and an unfamiliar extension gets fewer clicks at the same position. Lower engagement is a real business outcome regardless of whether it is a ranking input.
  • Practical friction. Form validators that reject anything that is not a familiar extension. Autolinkers that fail to detect the URL in social posts. Support staff who mishear the name over a phone. None of this is search, and all of it costs traffic.
  • Inherited history. This one has nothing to do with the extension and everything to do with the specific name. A dropped and re-registered domain carries its old links, its old blocklist status and its old associations, and no extension protects you from that.

Why most studies of this prove nothing

Every few months someone publishes a comparison showing that sites on one extension rank better than sites on another. Almost all of these are confounded, and it is worth understanding how, because the same error appears in expert analysis and in sales decks.

The populations are not comparable. Established organizations with budgets, brands and decades of accumulated links are overwhelmingly on legacy extensions. Newer extensions attract newer projects, smaller budgets, speculative registrations and, at the cheap end, outright abuse. Comparing average performance across those two groups measures who registered, not what the extension does.

The samples are also biased by construction. If you sample from what already ranks, you have conditioned on the outcome. If you sample from a registry's zone file, you are including enormous numbers of parked and non-functional sites. And the metrics used are usually third-party authority estimates, which are themselves derived from link data that reflects the same population differences.

A study that answers the mechanism question has to hold everything else still — the same invented string, the same content, no links, no history, no brand on any of them — which is why I built my own test that way rather than sampling live sites. Even then the honest reading is narrow: it tells you about discovery and indexing behavior for new, unknown hosts. It does not tell you what happens to an established brand, and anyone extrapolating from either kind of study to a mature site is overreaching.

Choosing an extension, in practice

My advice has been consistent for a long time and it does not involve rankings.

If a sensible legacy extension is available or affordable for your primary market, take it, because it removes an entire category of friction you would otherwise spend years explaining. If your business is national and stays national, the country code is frequently the better choice and it carries a genuine targeting benefit. If you are choosing a newer generic extension, choose one with a meaningful connection to what you do, check the extension's abuse reputation before committing, and expect to do more early work to establish the site than you would elsewhere.

What I would not do is buy a keyword-laden domain on any extension expecting the keyword to rank you. That was a real effect in the early 2000s and it has been discounted since. What still helps is indirect: if you build an actual brand on the name and use it as your business name, you naturally accumulate links whose anchor text matches the phrase, and that is where the remaining benefit comes from. A parked name with no brand behind it does nothing at all.

The test itself, and what each extension did, is set out in my CircleID piece, Top-Level Domain Bias in Search Engine Indexing and Rankings. Domain background checks run through DNAccess.

Frequently asked questions

Does a .com rank better than a new gTLD?

Not because of the extension itself. Search engines have stated that generic extensions are treated equivalently and that keywords in the extension are not a ranking signal. What differs is everything traveling alongside the extension: legacy extensions hold a far higher proportion of established sites with brands, history and links, and unfamiliar extensions face slower discovery, weaker user trust and more practical friction. Those produce real performance differences that are not a ranking adjustment.

Do country code domains help with local rankings?

Yes, and this is one of the documented cases. A country code extension is used as a geographic targeting signal, which helps within that territory and constrains visibility outside it. If your business genuinely serves one country, that trade is usually worth making. If you intend to expand, understand that you are choosing a structure you will later have to work around, either with a separate global domain or with subdirectories under one, and migrations of that kind are expensive.

Do keywords in the domain extension help?

No. A word appearing after the final dot has not been shown to function as a ranking signal, and search engines have said as much directly. The reason keyword extensions sometimes appear to work is that companies choosing them often build a brand around the whole name, and the resulting links carry anchor text matching the phrase. That is the same indirect benefit any brandable name produces, and it comes from the brand rather than from the extension.

Why do new extension sites seem to index more slowly?

Because indexing is a decision made under resource constraints using whatever signals exist, and a new site on an unfamiliar host has very few. The crawler has no history, no links and no brand recognition to work with, so discovery is slower and fewer pages are retained. It is experienced as a ranking problem and it is usually a cold start. The response is the work that establishes existence: links from frequently crawled sources, clean sitemaps, and unambiguous entity information.

Should I move my site to a different extension?

Rarely, and never for ranking reasons alone. A domain migration is one of the highest-risk projects an organization can run, and even a well-executed one carries a period of disruption while signals consolidate. Move when there is a business reason — a rebrand, an acquisition, resolving a trademark problem, or an extension whose reputation is actively damaging your email deliverability. Then treat it as a full migration project with a complete URL inventory and a redirect map, not as a domain change.

Does an extension's spam reputation affect my site?

Directly in email and security systems, indirectly everywhere else. Mail filters and security products maintain their own reputation data, and an extension with concentrated abuse can mean your messages land in spam folders regardless of how legitimate your site is. Browser and security warnings work similarly. For search specifically the effect is weaker and harder to isolate, but the deliverability consequence alone is enough reason to check an extension's reputation before building a business on it.
Top