Build it as an exact match domain

What an exact match domain is

An exact match domain (EMD) is a domain whose second-level label is character-identical to a search query with the spaces removed: cheapcarinsurance.com for "cheap car insurance", dallasroofing.com for "dallas roofing". The match is to the query — the string a person types into a search box — not to a product or a brand.

A partial match domain (PMD) adds something to the keyword, so it is not a character-for-character match: cheapcarinsurancequotes.com. A generic or category domain names the category itself — Insurance.com, Hotels.com — and earns its value from category authority and type-in traffic rather than query matching. A brandable domain is coined and describes nothing. An EMD can be generic too: carinsurance.com is both. But most EMDs are longer multi-word phrases that are neither: query strings with a dot-com on the end. The reason anyone cares is a fourteen-year-old argument about whether that string makes a page rank.

The September 2012 EMD update, and what it actually did

On 28 September 2012, Matt Cutts — then head of Google's webspam team — announced the update on Twitter in one sentence:

"New exact-match domain (EMD) algo affects 0.6% of English-US queries to a noticeable degree." (Search Engine Roundtable)

Three things get lost in the retelling. First, the number: 0.6% of English-US queries — a small, targeted change, not a purge. Second, the qualifier Google used everywhere else — low-quality. Search Engine Land described a "small" algorithm change intended to reduce low-quality exact-match domains from ranking so highly. The intent was never to demote keyword domains as a class; it was to stop a thin site outranking a better one purely because its domain matched the query. Third, Cutts said the change was unrelated to Panda and Penguin.

The most revealing part came a year earlier. In 2011 Cutts said Google had "been thinking about adjusting that mix a little bit and sort of turning the knob down" on the exact-match-domain benefit. Google was confirming a benefit existed and that it was deliberately reducing it. Anyone arguing that EMDs never helped at all is rewriting history.

What Google says now, and who says otherwise

The position has hardened since and is now stated flatly. John Mueller, in December 2023, answering someone weighing a keyword domain for a term with 10,000+ monthly searches:

"A keyword domain name is not going to give you any recognizable SEO advantage on Google." (original comment)

He has been consistent on keyword top-level domains too. In 2020: "You don't get a special bonus like that from having a keyword in your top-level domain." In November 2025: "There's absolutely no SEO advantage from using a .music domain" (Search Engine Journal).

Now the caveat, because quoting Google uncritically is not analysis. "No recognizable SEO advantage" is not the same sentence as "the domain string is ignored," and nobody outside Google can verify which is true. What is verifiable is that in fourteen years Google has never told anyone to buy a keyword domain for ranking reasons.

Against that sits a steady supply of posts claiming EMDs still work in 2026. Apply one test. Every source we could find making that argument is published by a domain registrar, an expired-domain tool vendor, or an SEO agency — organisations that earn money when the answer is yes. None publishes an inspectable methodology, and no search engine, academic study or controlled test supports the claim. That is not proof they are wrong. It is a reason to stop treating them as evidence.

Ranking, anchor text and trust: three claims under one label

Separating these three mechanisms turns a religious argument into a decision.

A direct ranking benefit. The algorithm boosts the page because the domain matches the query. Google says no, and no independent, replicated study since 2020 says otherwise. Treat it as either not real or too small to plan around.

An anchor-text benefit. People linking to a site very often use the domain as the anchor text, so if the domain is the keyword, exact-match anchors arrive without anyone building them. Anchor text is a long-established ranking input, which makes this mechanically coherent rather than magical — but it is second-order, and no published source quantifies its size.

A click-through and trust benefit. A searcher scanning results for "dallas roofing" sees dallasroofing.com and reads it as relevant. That is a claim about human perception, not an algorithm, and it is the most persuasive of the three. It also has the least behind it. No published study compares branded domains to exact-match domains for click-through rate in organic results. General CTR research exists — Authoritas, BrightLocal — but none of it makes that comparison.

One genuinely useful data point exists, and it is local-specific. Whitespark's 2026 Local Search Ranking Factors report, a survey of 47 local SEO practitioners, places "Keywords in Domain Name" at #71 in the local pack list, score 97. "Keywords in GBP Business Title" ranks #3, score 223, behind primary Google Business Profile category (227) and proximity of the address to the searcher (225). Practitioners believe the business name matters enormously and the domain name barely at all — survey opinion, not measured weights, but the closest thing to a consensus that exists.

EMDs and AI search

The plain finding: there is no published evidence that a keyword-matching domain helps or hurts citation in AI Overviews, ChatGPT or Perplexity. Not evidence that it does, not evidence that it does not.

The largest citation study located — OtterlyAI's AI Citation Economy, over a million citations across ChatGPT, Perplexity and Google AI Overviews in January and February 2026 — does not address domain names at all. Community platforms took 52.5% of citations against 47.5% for brand domains, and the barriers it identifies are technical and editorial: robots.txt, CDN restrictions, JavaScript rendering, content structure.

A reasonable inference is available, and it deserves labelling as inference. Answer engines surface entities — named things many independent sources write about. A domain string is not an entity; a brand can become one. If that holds, the shift to AI answers strengthens the case for a brandable name. No study confirms it.

Trademark risk: the UDRP three-part test

Registering a generic phrase is a commercial decision. Registering one that contains someone else's brand is a legal one. The mechanism is the Uniform Domain-Name Dispute-Resolution Policy (UDRP) — arbitration, run principally by WIPO, through which a trademark owner can have a domain transferred away without going to court.

Under paragraph 4(a), a complainant must prove all three:

1. "your domain name is identical or confusingly similar to a trademark or service mark in which the complainant has rights"
2. "you have no rights or legitimate interests in respect of the domain name"
3. "your domain name has been registered and is being used in bad faith"

The conjunctive wording in the third element is what domain investors most often get wrong: the policy requires registration and use in bad faith, not one or the other. A domain registered in good faith before a mark existed generally cannot satisfy it. Paragraph 4(b) lists circumstances evidencing bad faith, and one lands squarely on EMDs — using the domain "to attract, for commercial gain, Internet users" by creating a likelihood of confusion with the mark. Paragraph 4(c) sets out the defences, including a bona fide offering of goods or services.

The practical read: a generic EMD such as bestcoffeegrinders.com is low risk, because nobody holds rights in the phrase. An EMD embedding a mark is high risk, and developing it does not cure the problem — a developed affiliate site earning from a mark's traffic is precisely the 4(b)(iv) fact pattern.

When an EMD is the right call

The surviving cases share one property: the name earns its keep as clarity, not as an exploit.

• Local services with a city-plus-service name. omahagaragedoorrepair.com describes the business honestly, works on a van and a yard sign, and survives the radio test. Whitespark puts geographic keyword relevance at #2 in local organic factors, score 190 — but that is the site's relevance, not the domain's.
• A narrow niche where the query names the product. If the business does one thing and the query describes it in plain English, the EMD is honest labelling.
• Affiliate and comparison sites with no intention of building a brand — though Google's thin affiliation policy targets that category directly, so the domain is the least of the risks.
• Defensive registration alongside a brandable primary, redirecting to it.
• As a resale asset, where the buyer is a business in exactly that vertical.

The resale profile is unusual. A brandable has a wide, shallow buyer pool; an EMD has a narrow, deep one — real money to businesses in that exact vertical and geography, near zero to everyone else. The ceiling is what a single small business will pay. For scale, the domain aftermarket recorded 144,700 sales totalling US$185 million in 2024 per NameBio, an average around $1,280. Long multi-word query strings are absent from published top-sales lists — an observation about what is missing rather than a measurement.

When an EMD is the wrong call

• You intend to build a brand. A descriptive query string is generally not registrable as a trademark, is hard to remember, and caps the ceiling of the business. Mueller's own framing is that brand recognition beats keyword domains long-term.
• You need editorial links. Reporters do not link to cheapcarinsurancequotes.com. Every PR conversation starts from a deficit.
• You will ever expand beyond the keyword. dallasroofing.com cannot take you into Fort Worth or into siding without the name becoming a lie.
• The phrase contains a trademark. That is not a marketing problem.
• You are buying it because it is cheaper than doing the work. The 2012 update existed to stop a matching string substituting for quality.

The worst version deserves naming on its own: a hyphenated, non-.com, multi-keyword EMD. Hyphens appear in EMDs almost exclusively because the clean .com was taken. Users mistype them, they cannot be spoken aloud unambiguously, and they read as machine output. And buying dallasroofing.net because the .com is gone means paying a keyword premium for a name that donates its type-in traffic to a competitor, with no offsetting bonus from the extension. All of the branding downside, none of the upside.

Common mistakes, and what to do with an EMD you own

Over-optimised anchor text. The mechanical anchor advantage becomes a liability the moment the owner also builds exact-match anchors. A site on cheapcarinsurance.com whose backlink profile is dominated by the anchor "cheap car insurance" has a distribution no natural site produces — historically the pattern Penguin addressed. An EMD needs more brand and naked-URL anchors than a brandable site would, not fewer.

Registering the same pattern across many cities. Google's spam policies name "multiple domains targeting regions or cities" as an example of doorway abuse. A network of near-identical city EMDs is a documented policy violation, not a scaling strategy.

If you already hold an EMD, the useful conclusion is this. Stop treating the name as a ranking asset and start treating it as a positioning asset. It tells a visitor instantly what the site is for, which is worth something on a business card, on a van and in a search snippet. It will not rank anything on its own, and it will not substitute for service pages, links, reviews or a real operation behind it. Where the name describes a real business, keep it and build the business. Where it is a query string attached to nothing, build something genuinely useful on it or sell it to the one buyer for whom it plainly describes what they already do.

Frequently asked questions