Exact Match Domains an Unfair Ranking Signal?
One question I’m sometimes asked by people is about whether or not they should choose a domain name that includes the name of their business or brand, or if they should use keywords within a domain name to make it easier for them to rank for those keywords in Google and the other search engines. I often explain that while it may help them ranking for the phrase chosen if they use a keyword domain (often referred to as an exact match domains, or emd), that I usually prefer domain names using a brand, and that the best domain names tend to be somewhat short, memorable, and easy to spell, with emphasis on the “memorable.”
I have seen a lot of discussion on the Web about keywords in domain names, and a number of people discussing their experiments with exact match domains, and how those may help a site to rank for terms used in the domain name. The following video was uploaded at the Google Webmaster Help Channel this past March, with the Head of Google’s Web Spam team, Matt Cutts answering the question, “How would you explain ‘The Power of Keyword Domains’ to someone looking to take a decision what kind of domain to go for?”
A Google patent, originally filed in 2003, and granted today (with Matt Cutts as one of the listed inventors) describes this problem in more detail and provides some ways that Google could potentially act to lessen the value of keywords included in domain names (an exact match domains) by recognizing when queries are commercial in nature and using a different ranking algorithm for those queries that might lessen the value of domains with keywords in them. As Matt noted in the video:
We have looked at the rankings and weights that we give to keyword domains and some people have complained that we’re giving a little too much weight for keywords in domains. And so we have been thinking about adjusting that mix a little bit and sort of turning the knob down within the algorithm so that given two different domains, it wouldn’t necessarily help you as much to have a domain with a bunch of keywords in it.
The patent provides a little more context surrounding the issue, describing the use of keywords in a domain, or in exact match domains, as an effort to “trick” a search engine. From the video, it sounds like Matt empathizes a little more these days with site owners who are tempted to include keywords in domain names to help their sites become more visible in search results. Here’s the language from the patent on the problem that it attempts to solve:
In other situations, a company may attempt to “trick” the search engine into listing the company’s web site more highly. For example, if the search engine gives greater weight in ranking results to words used in the domain name associated with web sites, a company may attempt to trick the search engine into ranking the company’s listing more highly by including desirable search terms in exact match domains names associated with the company’s listing.
As an example, assume that company A sells laser printers. Company A may attempt to use a domain name that includes the words “laser printers” so that a search engine may rank the company’s listing more highly. As a result, a person searching for laser printers may not be presented with an unbiased set of results.
The patent is:
Systems and methods for detecting commercial queries
Invented by Amit Singhal, Matt Cutts, and Jun Wu
Assigned to Google
US Patent 8,046,350
Granted October 25, 2011
Filed: September 24, 2003
A system processes user queries. The system may generate a list of query patterns of a first type. The system may also receive a user query and determine whether the received query is a query of the first type based at least in part on the list of query patterns.
More than Commercial Queries and More that Exact Match Domains
The patent describes a number of approaches that could be used to identify commercial queries, and tells us that when a query is non-commercial, it might be processed in one way, and when it is commercial it might be processed in another manner that helps to “ensure that a person is provided with an unbiased set of results.”
The processes for determining whether queries are commercial or non-commercial may use an automated process, a manual process, or a combination of both, to find “query patterns” that can be used to match with queries typed into a search box to choose which algorithm is used.
Interestingly, in a paragraph near the end of the patent’s description, we also find this sentence which extends this process to include more than just commercial queries:
Moreover, while the above description focused on detecting commercial queries, implementations consistent with the principles of the invention are equally applicable to detecting other types of queries, such as queries for geographic information, navigational queries (e.g., a uery of “ibm” is likely looking for IBM’s home page), time-based queries, news-related queries, natural language queries, queries involving proper names, etc.
It’s also quite possible that in addition to giving less weight to exact match domains, the ranking algorithm used when a commercial query is identified may also look at other possible signals as well.
Identifying Commercial Queries
The patent describes a number of possible steps that it might take to identify commercial queries.
The first step may be to obtain a list of user queries, and it might limit that list to keep it manageable. An example from the patent tells us that it might “retrieve those stored search queries that occur at least once per 100 million queries.” That could potentially limit the list to a few million or billion queries.
The next step might be to collect a list of phrases or keywords of interest to advertisers or webmasters or both. That can include phrases and keywords used in advertising or phrases/keywords used in meta tags.
A list of domain names that contain 2 or more hyphens might be gathered as well. We’re told in the patent that:
It is very common to see domain names that include a single hyphen, but when two, three, or more hyphens are present, this is often an indication that these domain names are associated with companies that are attempting to trick search engines into ranking their web pages more highly.
Similarly, Google might create a list of hostnames (subdomains) that it finds that contain more than a certain number of hyphens. The Microsoft paper, Spam, Damn Spam, and Statistics (pdf), describes its authors’ observations concerning the use of heavily hyphenated hostnames in web spam. Google might collect a list of hyphenated hostnames during a crawl of the Web.
Google might watch manual and automated rank checking from companies to identify terms and phrases that those companies may be competing for against other sites to identify competitive queries.
A list of “short-circuit words” or words and phrases likely to be targeted by advertisers might be put together by monitoring queries received at the search engine, through experience with commercial queries, or by manual evaluation.
Processing Exact Match domains or Commercial Query Candidates
The lists of user queries, domain names, and host names might be processed in a number of ways, such as:
Removing stop words, digits, punctuation, etc. “For example, for the domain name “buy-credit-cards-online.com,” server may remove the hyphens and “.com” portion to leave the following phrase ‘buy credit cards online.'” In a query such as, “where can I find low apr credit cards,” the “where can I find,” might be removed to leave the phrase “low apr credit cards.”
An n-gram analysis of the list of domain names and host names might be performed to find combinations of words found in that list that tend to show up frequently.
For example, assume that the domain name list includes the domain name “buy-cheap-credit-cards-online.com.” Server may form the following exemplary n-grams for this domain name: “credit cards,” “buy cards,” “cheap cards,” “buy credit cards,” “cheap credit cards,” “buy cheap cards,” “buy card online,” “cheap cards online,” “credit cards online,” “buy credit cards online,” “buy cheap credit cards,” “buy cheap credit cards online.” Other n-grams may also be formed.
In addition to performing that kind of analysis on hyphenated domain names and hostnames, this process might also be performed on user queries and identified competitive queries, and terms or phrases that appear both within those lists and the domain/host lists might be identified (as “intersecting terms.”)
A set of heuristics or rules might be applied to those terms or phrases in lists where the queries and domains/subdomains intersect. For example, these rules could pull out any terms that might include two or more words, and a query occurs 5 or more times in the intersecting lists. Or three words long if the query occurs 2 or more times.
Other queries that aren’t identified through an intersection analysis, or through one of those heuristics might be identified as commercial if they include words identified as a short-circuit term. Assume for example that “hotel” is in the list of short-circuit words and the phrase “book hotel” wasn’t on one of the other lists. It may be identified as a commercial query since it includes the word “hotel.”
Other queries that weren’t identified as commercial through one of those processes might be sent to an ad server to see if it triggers a certain number of advertising related items such as ads or sponsor links or featured links, etc.
For queries that weren’t identified as commercial through any of those processes, the search engine may return a number of documents on a search for that query and examine those documents to see how commercial the documents might be. We are told, for example, that pages that target commercial terms might be more likely to include many keywords in their meta tags. (It’s possible that may have been more true back in 2003 when this patent was originally filed, but there are probably a good number of other signals that could be used to determine how commercial a page or site might be.)
We are also told in the patent that the search engine might also look at synonyms of the query terms and stems (or versions of the words included that go down to their roots – for example, “walk” is the stem of “walking.) The types of analysis described above might be performed with those synonyms or stems.
The patent notes that the processes above are illustrative examples, and provides more details for a number of them as well as a few alternatives and possible ways to score how “commercial” a query might be.
We don’t know if Google is using this approach to keywords in domains, or exact match domains, but if not, it seems like they potentially might use it or something like it from Matt Cutts’ statements in the video I included at the start of this post.
The process in this exact match domains patent focuses upon queries that Google might consider to be “commercial,” so it’s possible that keywords in domains might work better with non-commercial queries than commercial ones if Google follows this patent.
The patent was originally filed in 2003, and it’s possible that Google might look at other signals as well that aren’t described in this exact match domains patent, but I think it provides an interesting look into some of the assumptions from Google about keywords within domain names.
Lat Updated May 22, 2019
120 thoughts on “Google’s Exact Match Domains Patent (Detecting Commercial Queries)”
Even with Google implementing the above it will be extremely difficult to enforce as there are a lot of people out there with excellent sites with great content that use exact match keyword rich commercial domains therefore it would be unfair to penalize these guys for this.
The keyword in domain name would always be a good idea. If I would be asked for the characteristics of a good domain name, I would always consider: Short, Easy to remember, Contains keyword, Brandable, Dot com domain, Hard to mistaken & Hard to misinterpret.
There are several things you can do to get your final domain name ideas: Use domain name idea generators like: MakeWords.com or Nameboy.
Implementing the above by Google will not be fair.
I have already seen some shift in local SERPS in my area for the keyword (Glasgow SEO) and its variants, so I imagine that this is already at work to some extent.
Although Google may start to remove the super powers from EMDs will it be able to do this to the correct degree? For example, if you have an EMD you can spam exact match anchor text at it all day and not get penalised as it is just the domain name you are linking with. Is there any mention about this patent looking further than the domain name to decide rankings for a term? And what if I actually register my company as a Exact Match Keyword LTD. Much like the cheapflights.com guys have done.
For the sake of the quality of the web I am very happy about these revelations but being the owner of a few excellent EMDs I am also a bit bummed. Maybe I should conduct a little test with an EMD and a branded domain.
I have always been of the belief that a domain name with the keyword in it would be more beneficial, especially in a niche area. The “brand-able” examples you gave are all one word URL’s, eg: yahoo, digg. How much weight does a single word URL have opposed to a long URL?
I would say it’s best to have both brand and product in domain if it makes sense to do so and is not too long. If not my next obvious choice is to go strong on branding and not repeat with domains what happened to Yellow Pages business names (AAA+++ Aardvark…etc). It’s certainly interesting to see Matt on the patent docs and will be looking out to see if they solve the problem of commercial searches versus true brands in the future.
I really hope they turn the “knob down” with EMDs. It’s a little ridiculous right now how much SERPs are dominated by them. However, like Ross said above, being able to use exact match anchor text with impunity already gives them an advantage that I’m not sure Google can overcome right now. In the future as they move towards rel=author and social signals, EMDs might not matter as much.
I still don’t see any particular advantage in exact-match domains. This remains more a myth of perception than presentation. Every competitive query I have looked at where exact match domains supposedly rule the day turns out to be diverse.
I think this falls in the bucket of Google hates “anything easy”.
It is reasonably easy to get EMDs to take over page one results in a few weeks for what I would consider medium level searches (ie – <1000 monthly exacts).
Sort of sucks that a few bad apples can spoil the party for the rest of us. 🙂
I agree that they likely will not kill this for everyone as I am sure there are as many (or more) good quality sites on EMDs as spammers.
It seems Matt believes the world revolves around Google and everything companies do online have to do with placing in their (currently) top-rated search engine.
By the above descriptions, I resent the fact that some of the domain names we registered and developed into comprehensive sites would be considered by Google as “gaming” their engine, based solely on the fact that the words that comprise the domain may be a popular search phrase.
Matt, have you heard of DESCRIPTIVE? Our company, as others, believe descriptive terms such as “laser printers” in a domain name are helpful to CUSTOMERS who we’d like, at a glance, to see what we’re offering.
Many companies like ours advertise our websites on TV, billboards and in print, and having descriptive terms in the domain name helps potential customers know what we’re offering.
This “Google is the center of the universe” is getting nauseating.
Should the first person that manages to register or acquire an exact match domain be given some kind of search ranking bonus because of their domain name, even if there are possibly more relevant or better quality pages covering the same topic? If someone starts a site with an exact match domain, and then they decide to change the focus of their business or marketing, or even just the ideal keywords to use for their site, does that domain name choice then become a detriment rather than a benefit?
I’m not sure that this is a question of “enforcement,” rather than a matter of not giving as much value to the ranking of a page when a certain domain name has been chosen that includes keywords within it. Those excellent sites with great content should still rank well without an exact match bonus.
We actually ran a small test a few years ago on 100 keywords and over 1000 sites and found that, for common keywords, EMDs mattered little. It was only where they keyword also was a brand that EMDs mattered. The key here is about good branding.
I agree with you that giving more importance to the domains that have exact keywords in them isn’t fair for other sites. I don’t know about the future but right now I can see that most of the SERPs are dominated by EMDs. Whatever I search for there are 40-50% sites with EMDs there. So as far now, search engines do seam to favor the sites with exact match domains. But I completely agree with you that just having the keywords in the domain name should not give you an extra benefit over the other sites. And by seeing the current results I can say that search engines give a lot more weight to the EMDs.
It’s hard to tell whether or not Google has implemented the approach in this patent or not at this point, but it’s possible that they might have. I did have someone I know show me a couple of sites that he created a couple of months ago that looked like they were getting a benefit from exact match keywords, but I’m not sure that the keyword phrases he used were very commercial.
I would imagine that like with any algorithm that Google might implement, there would definitely be some testing going on to try to get an idea of how much weight a domain name might pass along, and what would have to be done to “present unbiased results” (to use words from the patent) to searchers.
The patent pretty much focuses upon signals that they might look at to identify commercial queries rather than the steps they might take to remove the ranking value of keywords within domain names, or other signals to rank search results.
The mention within the patent’s description that Google might use similar approaches to identify navigational queries, or geographically related queries was interesting as well. I think Google has done that in a number of different ways, for example identifying named entities within queries, so that they might associate specific sites in some instances with certain queries, so that someone searching for [nike shoes] might see the Nike website as the first result, or a query for [hilton london] might include a link to the Hilton hotels website as the first result.
Websites that use the targeted keyword in the domain name often get high rankings, but make sure you concentrate on every aspect of possible ways of online marketing methods, because it is been over 6 years now and I am still waiting for an example where a website ranked well only because of the domain keywords. I am not denying its benefits, but it is just part of whole business game.
Ian has a very good point on the length factor, not only in case of domains but I am really interested in knowing that, if the weight of the primary keywords gets reduced if the whole domain name, url or the anchor text becomes longer.
We are not in favor of penalizing the sites with EMDs, we just want that the search engines do not give too much importance to the keywords in the domains.
Interesting that they may use separate algorithms in order to return results for certain types of queries. It makes sense to handle different types of queries through a different set of analyses, and this also explains some of the results pages we’ve been seeing which are related to ecom-type queries. Thanks for the great summary of the patent!
By longer I mean if the primary keyword is “bicycle” how much it can affect the rankings for the primary keyword if someone uses “xyzbicyclesindia. I hope Ian is also asking the same thing. But I am really unsure as well, as to how much importance it holds in the search engine’s ranking signals.
Thanks for clearing this out Bill. I would like to ask a question that I wanted to understand more, since its related to this topic, I read somewhere, maybe it was Rand or someone else from SEOmoz staff explaining that why SEOBook.com ranks higher than SEOmoz (it was above SEOmoz at that time) when we search for SEO. He explained that that is because SEO and Book are both meaningful words whereas SEOmoz has moz in it which is not a word. So, does that also affects the rankings. Does Google take into account the meaningfulness of the domain name. I mean if the domain contains meaningful words along with the primary keywords would be favored more than the domains containing non-meaningful words in the domain. I would like to know what you think?
I would have linked to the source where I read that, but right now I am unable to find the source.
Interesting. It’s hard to tell whether this is already in place, to be honest I haven’t seen much difference recently. I guess only time will tell, and if they do ‘turn this on’ soon, hopefully there is not too much collateral damage as a result. Granted there are many EMDs out there which have been created specifically with the view of gaming the system, but there are also many companies whose business name coincidently also happens to be a highly searched term – many of these businesses chose their name before they even knew about SEO.
Good questions. I’ve always been a little wary about how much value an exact match domain does impart upon a website. A lot of the value they might bring could be attributable to links pointing to a page or pages of the site that use the domain name. There really isn’t much out there on the Web directly from the search engines that says, we look at keywords in domain names and use those to rank pages. I’ve looked for those types of statements. If someone knows of any from any of the search engines, I’d appreciate being pointed to them. How much weight does any URL have, regardless of its length?
My biggest concern is that a race to register a domain name that uses certain keywords shouldn’t provide such a strong signal that other more relevant sites for those keywords shouldn’t be able to overcome it.
I think Yahoo and Amazon were both great domain names when they were chosen, and there are no keywords in those. I love flickr as a domain name, even though its spelling might confuse people.
What’s unfair is that there are only so many choices of domain names to go around, and just because someone registered one first with certain keywords in it doesn’t mean that it should outrank sites that are more relevant, or provide more useful and helpful and meaningful information. If getting that domain name first is the thing that makes a difference, then that’s not really all that fair.
I lean more towards using terms in domain names that can help with branding than with keywords that you might want to be found with as well. I see the value in chosing a domain name that might include both, not only from a search stance, but also from helping people who see the domain name understand what they might find at that domain. But if it came down to keywords or branding, I would probably choose branding most of the time.
It was interesting not only to see Matt’s name on the patent, but also that this was something he had been involved with and thinking about for such a long period of time as well.
When there are so many other potential factors and signals that search engines might look at, it’s hard to tell how much impact a single factor such as a domain name might have. It’s also possible that very little of that might come from the domain name itself, but could be attributable to other things, such as the use of the domain name on the pages of the site at that domain, links pointing to the pages of the domain, possibly some type of entity association between query terms and the pages of the domain, and others.
I agree with you that as Google broadens the range of signals that it might look at, that any that might be weighted too strongly will have less of an impact. Especially if those range of signals start measuring things like credibility, authority, and other attributes that aren’t as singlemined.
I’ve looked at a lot of sites that use exact match domains and have seen a number of reasons why those sites may rank as well as they do based upon other factors.
I’ve also seen pages that rank very well in some situations based upon a very limited number of factors that are sometimes pretty hard to isolate and point at, and might be easily misinterpreted as being caused by an exact match domain.
I’ve considered myself pretty skeptical of the power of exact match domains over other signals, and believe that the best approach to SEO doesn’t rely too much upon any one signal.
Hi SEO by the Creek
I’m a little closer to the creek these days than I am to the shore. 🙂
I beleive that the patent focuses upon commercial terms rather than noncommercial terms in giving less weight to keywords in domain names for a reason The assumption behind in the patent seems to be that spammers are much more likely to focus upon using keyword terms for more commercial terms that might bring them more money for their efforts.
I think that Matts more recent statements about exact match domains (such as those in the video) shows a recognition that many website owners that don’t have an intent to spam struggle with the question of whether they should choose a domain name that might help them be found more easily as opposed to one that might help them build a stronger brand, and the intent isn’t to spam as much as it is to be found relevant by the search engines.
Thanks for sharing your data about your experiments, and your findings.
I think sites that do a good job of creating a brand can get a real benefit from Google associating their brand with one or more query terms, and may get a boost in search results for that term or terms based upon that kind of association.
Appreciate your update on your experiment as well.
That’s the kind of healthy skepticism that I respect, and I agree with you that I would definitely do a lot more than just rely upon some keywords in a domain to help make a site show up in search results for people interested in what it has to offer.
Google may or may not be giving more relevance value to pages on a domain that use keywords within the domain, but as this patent hints at, something like that could disappear overnight.
Thanks. One of the impacts of the Google infrastructure update referred to as “Big Daddy” seemed to be to make it much easier for Google to use ranking algorithms in a modular manner, where they could switch in and out algorithms with the pressing of a button. Want an algorithm that might treat tech blogs and food blogs differently, because you’ve introduced recipe smart snippets, just rewrite the algorithm, test it, and then press the button and make it live.
We known that Google’s been applying different ranking algorithms for different types of sites for years. Since this patent was originally filed back in 2003, and describes in a lot of detail how a commercial intent might be associated with some queries, I think it shows us some of the earlier history of how intent might be considered in showing search results.
I’m not sure how much of a difference EMDs are having these days either, but I’ve seen a number of blog posts complaining about them over the course of the past year.
Good points about brand names and businesses that chose to incorporate terms within their names that not only were brandable, but also descriptive of what they did, or do.
Thinking about that, it’s funny sometimes how quickly those change. For instance, 3M started out as the Minnesota Mining and Manufacturing Co., and within three years or so the mining aspect of their business was forgotten as they abandoned the mine they were going to take abrasives from, and started focusing upon sandpaper.
I agree that I hope there isn’t much collateral damage to sites that might rely too much on EMDs, and I would recommend to anyone who might be that they beef up other aspects of their sites so that they aren’t harmed.
The problem is that we don’t know how much the exact match domains help sites rank for specific queries, or if things like links to the pages of those sites using the domain names are the kinds of things that tend to have the biggest impact.
When Google does some kind of entity association where it might associate a keyword phrase with a certain website, is that being done because those keywords might be included in the domain name? It could be though a good number of other things need to be happening as well for that kind of association to happen.
With domain names being relatively short compared to other things that a search engine might look at as a ranking signal, I’m not quite sure that I understand the point that Ian is making about the length of a domain name.
There are some times when I read a search related patent when I wonder if the people who wrote it ever owned a website before or ever tried their hands at ecommerce.
I’m personally a big fan of building a brand that doesn’t necessarily rely upon a search term (Amazon, Apple, Yahoo, Google, Flickr, etc.), but using a domain name that’s not only brandable but also descriptive can make a lot of sense as well (International Business Machines, for one very large example).
I don’t know that the Matt Cutts of today would have used the same language in the patent that he or his co-inventors used in 2003, but they might have.
Having said that though, I don’t think that a business should necessarily get a benefit because their business name or domain name matches up with a particular set of search results, regardless of their reason for choosing the domain name that they chose.
I don’t know if we’ll ever find the answer to that question, but thanks for explaining what you meant.
Another thing that Google has been working on is segmenting text when it finds in in a form like that. I would expect that you might include XYZ Bicycles India as a page title on the site, so that would help the search engine figure out where to make breaks in the text.
But does a longer domain name that includes a number of potential keywords mean that less emphasis and less relevance might be seen for each individual term? I don’t know. We also know that Google is trying to understand meaningful phrases when it comes across them, so “XYZ Bicycles India” might be determined to be a phrase and a named entity, and rank well for the phrase, but possibly not as well for the individual terms within the phrase.
To be honest i am not very technical but i have always beleived that to own/run a site with a relevant name to the product would be the most desirable out of all options, to be be punished for this would be crazy. Should the site contain quality and relevant content? Of course, the domain should compliment the content and both should work in zinc with each other
I don’t consider myself an expert, but I do believe that having your main keyword on your domain name is important!
Interesting question, and possibly one that would make a good blog post.
Both seomoz and seobook have been around for a few years, accumulating a fair amount of links, publishing original and thoughtful commentary and content centered around SEO, offering free and paid tools that people link to, and both have a considerable amount of traffic coming their ways. Those are definitely the kinds of things that can help them rank well for a pretty competitive term such as SEO.
Let’s say that keywords within a domain do matter, and that a decent percentage of links pointing to either site use the domain names for those sites. Does the use of those particular domain names, the inclusion of the site names in page titles and the content of their pages, and the links to “seomoz” and to “seobook” help either site rank well for the term “seo” or does Google just give them credit for the madeup compounded words? Does Google treat SEObook differently that SEOmoz because “moz” isn’t really a word?
Google will try to segment words that it might find in a domain name and within URLs when it can, and it’s possible that it will do the same thing when it finds words in other places as well. There are some Google patents that deal with text segmentation that might be appropriate to look at, as well as some that deal with decompounding compound words. Google’s Phrase based indexing may also play a role in how relevance might be determined for those once they are segmented or decompounded.
To try to understand how Google might treat each, I’d probably look at the following patents:
Methods and systems for selecting a language for text segmentation
Methods and systems for improving text segmentation
Both provide a way of thinking about how Google might attempt to separate strings of text into meaningful words, and I could possibly see the processes described within the patents breaking down SEObook into “seo” and “book” a little easier than breaking SEOmoz down into “seo” and “moz” since “moz” isn’t something that tends to occur independently as a separate word on the same kind of frequency on the Web that “book” does, and it may also be a little difficult for Google to determine its language.
But let’s say that Google manages to segment both so that it understands that “SEO” is a separate word within each. Google may try to determine whether or not each separated or decompounded version is a meaningful phrase, as described by one of the phrase-based indexing patents, and if it decides that “seo book” and “seo moz” are meaningful phrases, it might possibly treat them as being more relevant for those phrases than for the individual words within them.
But it is possible that Google will consider the inclusion of a keyword within a domain, and for both sites, the use of those words within their domains, and in links pointing back to them using the domain name, using a text segmentation process to identify those, though it might be tempered a little by looking at the terms as phrases rather than as individual words (much in the same way that “ice cream” means something different than “ice” and “cream” appearing on the same page).
I’m not sure that we need to worry too much about either segmentation or phrase-based indexing for either site though, because both Rand and Aaron have worked to optimize their sites, and even links to them, with those words joined together, and with them separated. For instance, if you look at the slideshare account for seomoz, you’ll see that they’ve segmented the domain name into two separate words: “seo moz.org”.
Both sites also have pages that are optimized for “SEO” independently of their names or domain names either, and the page that shows up highest for seomoz in rankings for SEO was “The Web Developer’s SEO Cheat Sheet | SEOmoz” on one search, and their SEO beginner’s guide on another search.
What this shows to me is that a good SEO won’t rely just upon the possible power of an exact match domain, or a domain that has an important keyword within its domain, but will also do what they can to rank well for the terms they care about in other ways.
It’s possible that there may be a benefit to include a keyword relating to your business or product in your domain name, and in your business name. But what if you offer multiple products with different names, or change your businesses objectives? I’ve seen things happen like a company selling storm windows and chosing an appropriate domain name (example washingtonstormwindows.com) and then decide that they would be better off focusing just on storm doors, and have to decide what to do with a now inappropriate domain name.
I think I need to stress that this approach isn’t aimed at punishing a site as much as it is not giving a domain name with commercial keyword terms in it more relevance and ranking credit than other sites that may offer similar services or goods but don’t have a domain name with keywords in them.
There can be a real benefit in having a domain name that includes your main keyword within it, especially since it tells people more about what you offer on your site when you see it.
The Google patent seems to focus upon treating sites with domain names equally, regardless of whether they have keywords in them or not, by finding a way (for at least queries that they think are commercial in nature), for the domain names with keywords within them not to get an extra benefit because they included those keywords.
On thing No one mention was that EMDs probably have a much higher clickthrough rate! And if your EMD gets a tons of clicks then won’t that have an affect on you rankngs?
You raise a really interesting point. Is is more likely that someone will click on a search result when the domain shown for a page contains one or more of the query terms that they used? I did write about some research that Microsoft did on different elements in search results for pages. They called the combination of title, snippet, and URL a caption for each page, and they tested different elements of those captions to see how they might influence clickthroughs.
I wrote about it here: The Influence of Search Result Listings (Captions) on Clickthroughs
The Microsoft paper can be found here: The Influence of Caption Features on Clickthrough Patterns in Web Search
Instead of just looking at the domain name, they also explored whether it made a difference in terms of clicks when query terms appeared anywhere in the URLs. The paper is written from the perspective of search engineers rather than SEOs, and It appears from one of the tables included in the paper is that when a query term appears within a domain name, it may receive more clicks.
I can’t recall seeing any other whitepapers that explore that in more depth, or discuss it in more detail, and I’d love to see more testing and results that might cover the impact of query terms in a domain and a URL upon clicks.
In my opinion, Google should not even put as much stock in alleged keywords in domain names as they might still do after implementing the patent. Why? Simply because a domain can be pretty arbitrary – what if a natural surname happens to be a “keyword”, like, believe it or not, “Miller”, “Smith” etc.? Would it not -from a search user point of view- be always best to look for the CONTENT in any of the subpages that might (or might not) fit the search terms? What if someone has, say “concrete” in their domain but builds, say, wooden yachts? Or gives “concrete advice” in, say the area of marriage counseling?
So tired of crappy exact match domains getting top results anyway. I know many of them have effort put into them, but I am pretty sick of best-flowers-for-your-girlfriend.net etc etc.
I’m always amused by the folks that insist an exact match domain isn’t easier to rank without having actually tried it themselves. It’s quite possibly the easiest test to try and pretty cheap too. Grab a handful of EMDs, post a page of content, link from your blog or somewhere and watch your first page ranking materialize. It doesn’t work 100% so you shouldn’t rely on one EMD and you won’t win in competitive markets, but if there’s an EMD with good search volume that you can grab you’d be silly not to. Having said that, I wouldn’t rely on the EMD working forever so it should be a secondary property with a branded property getting most of your attention for the long run.
yes this has been around for years, and certainly hasnt been implimented yet, as I note in my own market. Its crazy to do this anyway, what kind of branding can you give a site called, ecommerrce seo web design solutions serivce…exactly…none!!
All other things being equal why not attribute a few foot pounds of lift to the developer/enterprise that was savvy enough to register or secure a domain that unambiguously defines a website’s purpose?
Why shouldn’t a domain that conveys a clear message, one clearly aligned with the associated website’s purpose, be a +1 value signal in an increasingly noisy Web? Because or or anchor text or paid links or the other 196 variables do a better job? Hardly.
Brands are great if you can afford the expense of disambiguating your website’s purpose. However few companies have the kind of budget needed to transform a South American river into an ecommerce juggernaut.
Ambiguous website names, i.e. “brands in the making”, are make work projects for marketers and a slew of associated industries (SEO, PPC, etc). Forget the make work expense of branding and secure the “clearest message domain” money can afford: Hotels.com, Insurance.com, CreditCards.com. Then build a site worthy of the domain.
The brand dialogue then becomes what it should be, i.e., what others are saying about the quality or value of your service.
As soon as ICANN allows the general public to register a top-level domain with any suffix such as .brand or .keyword we’re bound to see more keyword stuffed domains. Change will bring new challenges!
The only reason that I can think of for Google to implement something like this is that they want the value of keyword domains to be transferred to ad $$$.
Think about it, even if you own creditcards.com and have a good developed site on it, you’ll still need to pay Google for traffic from the term ‘credit cards’ which you might not have had to earlier.
Sucks, but par for the course considering Google is evil and greedy.
Thanks for the explanation Bill. So are you planning a post on this?
Its a really tricky task as Google need to assess whether a change in ranking would improve the search results. One thing I did notice over the last year is that more and more EMDs are being launched and typically grabbing first page rankings for competitive terms. Over the past week in my niche I did notice that one of my own sites knocked off the #1 ranked site which had an EMD. The site however was a very poor site, had little content and contained no physical address so this would be beneficial.
The problem comes when a legitimate site is occupying a good ranking based quite heavily on the fact they have an EMD. Personally I think that websites that use an EMD for the sole purpose of improved ranking should have the benefits removed completely. This should at least level the playing field and stop websites being launched with a URL like ‘smallbusinesssearchengineoptimisationcompany.com’ even though their company name is something completely different.
Google trying to protect the dollar or it search looking to level the playing field would be a plus.
Using keyword for domain name would be a great idea.I think that we should take a short name for the domain. It would be one word or two words.
Hi SEO Novice,
Good points about surnames, and looking at the actual content to be found on web pages associated with a domain. I think we may see less and less reliance upon keywords found in domain names in the future.
I have no problems with sites that use exact match domains, if they are actually relevant for a query and provide quality content for what searchers might be trying to find. I don’t think that their domain name by itself should mean that they should rank higher than pages that provide more relevant results.
Regardless of the potential benefits of an exact match domain name, there are so many other signals that the search engines may be looking at that it doesn’t make sense just to treat a domain name as if it were a magic bullet. I’m not against anyone choosing to use an exact match domain, but I would definitely recommend doing everything else necessary to make a site successful as possible.
There are some businesses that have built successful brands around what might be considered keywords, such as International Business Machines.
And yes, it’s possible that something like this may have been implemented in some markets and not others.
Building a brand isn’t an exercise in SEO; it’s part of a business plan itself – deciding how your business is perceived by others in the actions that you take, in your interactions with others, in defining what makes your business unique and different, in creating a particular mindset about your business. It’s not a “make work” exercise, but rather a succeed at business endeavor. A company doesn’t have to be an economic juggernaut to build a successful brand.
All other things aren’t equal, and there are many other solid and meaningful signals that a search engine can look at that provide it with enough information to determine whether or not a web page is relevant to a particular query, such as a page title and the contents of pages.
Wouldn’t that be the same as building a brand?
We have yet to see how the search engines might handle those, but it should be interesting.
I’m not sure that a suffix will be seen as a relevance ranking signal if there isn’t a sufficient match between the choice of suffix and the content that appears on the pages of those sites.
You know, I would hope that the most relevant sites show up in the top results on a search for [credit cards] rather than the one which paid the most money for a domain did. I don’t see this as a grab on Google’s part for more advertising dollars.
I agree completely – level the playing field so that things like relevance, quality, popularity, authenticity, credibility, and usability become more important that choice of domain name.
I’m still considering a post, though my response to your comment might have been long enough to be a post on its own. It wouldn’t be about seomoz vs. seobook though, but rather how Google might segment a domain name when it contains multiple words, or what might be considered to be words.
It does seem to be an effort to level the playing field.
I am also a fan of brandable domains that don’t necessarily make use of keywords (your reply to Marcia) although, as you have mentioned, if one can be incorporated then it might enhance the brand if done wisely as you mentioned in your reply to Rowan.
In fact seobythesea might be a good example of a brandable name which incorporates a keyword, notably the first word in the name.
Interesting also to consider how the words in the domain might be ranked separately or in various combinations and permutations. In google.co.uk, seobythesea.com ranks on page 1 for [see seo]. Is this ranking aided more by the fact that the word [see] occurs in the content, also possible misspellings of back link anchor text and maybe even possibly because of the word [sea] in the domain?
hi friends am not an it expert but just hobby domainer
my question is that when you type some world or keyword in Google search bar ie (world map)it indicate About 286,000,000 results (0.09 seconds) what does is mean.Any word or key word
which return over 150 million results or as in case of world map 286 million result.if such word or key word is taken as a fresh domain, is it worth registering and it carry some value if it is exact key word for any search query.
i am new to your site but find some value able stuff.hope u can help me to sort out my query.i hold some domain such as shejunction.com .when entered in Google search bar (she junction) it return About 712,000,000 results.my question is, is it worth retaining and get it developed or put it on clearance to replace it with some thing else.hope u will help me to clear my mind.
When I came up with the name “SEO by the Sea,” the purpose really had little to do with attracting traffic to the site, and more to do with providing a descriptive name for an event that I was going to hold as a free conference on internet marketing, along the lines of a barcamp. It was held in Havre de Grace, Maryland, and I gave it name that told people clearly that it wasn’t in some huge conference center in some concrete jungle.
The site evolved into a place where I blogged about search, SEO, and search related patents and whitepapers, and it is an example of how a keyword can be incorporated into a brand.
The site ranks pretty well for [see seo] in google.com for me as well, though that wasn’t something I focused upon in anyway. I haven’t analysized that in the past, but may look into it now. It ranks even better for [sea see], to the point where it’s showing sitelinks for my site as well, and Google may be interpreting that query as a misspelling and an intent to look for my site specifically.
When you type a particular query into Google, it shows an estimate of the number of pages that might contain those words, or even synonyms for those words in many cases. But it’s only an estimate, based upon looking at a small percentage of results that might rank for those terms, and extrapolating from there.
I’m not sure that information is tremendously helpful by itself in determining the value of a possible domain to purchase. You might want to look at other things, such as how many results come back when you put quotation marks around the phrase, how many results you might see when you do an allintitle search, or an allinachor search or allinurl search (allintitle:world map, allinanchor:world map, allinurl:world map, allintitle:”world map”, allinanchor:”world map”, allinurl:”world map”).
This patent tells us that Google might not give as much weight to the words within a domain if it believes that those keywords evidence a commercial intent, and it tells us about a number of ways that it might decide if they do.
Of course, it’s also worth considering who your competitors might be if you decide to focus upon those terms as a main keyword for your site as well, and how difficult it might be to try to rank for that phrase regardless of whether or not you might potentially get some kind of boost for having the keywords in your domain name.
Hi Bill, very interesting – and generous of you to make such a detailed reply.
You’re welcome. Thanks for pointing out the rankings for [see seo]. It’s not something that I focused upon, mainly because it’s not likely something that people interested in learning about SEO or trying to find SEO services might use to find a site like mine. But it is pretty interesting nonetheless.
I wrote a post on Search Engine Journal some months ago about EMD ( http://www.searchenginejournal.com/exact-match-domains-this-is-why-they-can-still-rank/28891/ ), but now I don’t know what to think. The patent is dated 2003 and I still see exact match domains having a navigational boost… I guess google could act through these years: if it hasn’t done maybe it can’t…
The patent describes a process that doesn’t necessarily limit the value of exact match domains for all keywords, but seems to focus upon ones that evidence some kind of commercial intent. It does lay out a number of different approaches that could be taken either together or independently (possibly along with others that haven’t been included in the patent) to understand what queries and keywords might be commercial.
So what may work in terms of keyword domains may differ from one query term to another.
Having your main keyword in your domain helps… but for the reason that it is easier to get the keyword in your anchor text of links. Paid directories is a perfect example of places where you can get anchor text with your keyword in it from trusted sources that will only link to you with your company name.
There is definitely some value in including one of your main keywords in your domain (as long as that keyword continues to have some relevance to your site), but it’s as likely that you’ll get that value from the title of your main page if the keyword is part of the name of your business or organization.
I hope this isn’t getting too off-topic but it’s an interesting point about keywords in domain name (and/or business name) and how much weight is given to the anchor text, in back links from, say, business directories as Brian has mentioned, as a direct result of those keywords being present in the domain name.
Surely if I were to link to a site about SEO from a directory page which has SEO in the page title, only has other sites about SEO on the same page i.e. the whole page is about SEO, Google can work out that this must be a citation for SEO and therefore not need the additional signal of the keyword [SEO] in the anchor text. If so might that not lessen the advantage (in that particular circumstance) of having the keyword in the domain?
Under a straight hypertext analysis, having a keyword within anchor text pointing to a page should help that page seem more relevant for that keyword. Under a phrase-based indexing approach, it really helps if the content of the anchor text is in some way related to the content found on the page itself.
It’s hard to say whether Google might take in the context of a page that a link appears upon and assume that even if the anchor text within the link doesn’t seem relevant for the page being pointed to, that Google will assume that it has some association with what the search engine finds on the rest of the page. I’m not sure that they would. Under a reasonable surfer approach, the use of anchor text that doesn’t seem very relevant might mean that the link might not pass along as much PageRank and hypertext relevance.
I’m a big fan of exact match domain names, where they make sense. Of course they are only relevant for a small percentage of sites.
well, here’s the problem, my father, who’s 85 now, uses google to start with; whatever domain he wants to visit, he doesn’t type it into the address bar, he puts it in google, so that’s why google has to give priority to exact domain match …
I agree. There are times and places where exact match domains do make a lot of sense. I’ve created a few sites where that was true, and I wouldn’t go back in time and change them. Given that though, I’m still a fan of using a domain name that people will remember, and sometimes exact match domains can be pretty generic sounding.
Hi Bill, excellent discussion.
It appears that there is still a lot of head scratching. As I read, I wondered has anyone thought that Google is laying down something quite simple here?
Domination of Big Business by default on the web in the hands of a search engine deciding if YOUR content is commercial or not and whether YOUR content is large enough to compete with JOE BIG BRAND getting the nod first by default of their brand useage.
As others have mentioned keywords are important such that domains with them indicate the type of business offered and targeted cusomers find them because that is how targeted the customer searches these days.
There was a time on the internet that when a place like fast dudes dot com could be brandable but for what? Hence fast dudes VW repair dot com was born to stop the search madnesss and we could find their relevant site for what I was looking for. This was actually a good idea AND it allowed the SMALLER businesses to compete with the LARGER shops that repaired VW’s. The Larger shops have now realized that THEIR brand possibly was not that strong with the specificity of the competing smaller ‘mom & pop” EMD’s entering the picture and getting some business.
Obviously there is plenty of abuse to go around on all sides here.
It appears, IMHO, that Google with thee “ALGORITHM” wishes to skew the possible playing field back in favor of LARGER BUSINESS brands again including their own with all of the other social media gaining popularity over them. Google actually is beginning to be less “RELEVANT” dare I say yet still important “as a search tool”
Obviously the smaller players [ie. the 1%’ers >>me] who are struggling to make a buck online take that buck away from JOE Big Brand. So, Google can gain their dollars back by catering to them to assist with their keeping business in their pocket, How you might wonder? They pay Google money that I can’t pay. That makes them RELEVANT in Googles eyes. I hope nobody wishes to argue that one.
I say that the EMD’s should take their domains and ADD relevant content as best as they can and HOPE that they can still compete.
If anyone goes to Alexa and looks at who is ranked for what, they will see that NAMES do not matter much if you HAVE 220,000,000 links back to your site and 320,000,000 links within of “someone elses” content and your site has a name like Snortz dot com [didnt do a whois >> is it real?} and that name would not tell me a thing about their business. SO, does SNORTZ get bonus points for doing an IPO and raising cash for the programmers to get automated linking and free profiles with social media set up mean that the content is relevant content or does it mean THERE IS A LOT OF CONTENT? I think we all know that answer.
Seriously, so when people get concerned about the weighting of a domain with EMD’s where is the opposite end of this spectrum that SNORTZ dot com has all of this NON-relevant and NON-focused content linking that could blanket the world 200x just twittering around my you tube gaing a gazillion more links of content that when coupled with a search Algorithm that favors that scenario, Of course EMDs are seen as threat because Google has TO TAKE THEM INTO ACCOUNT because they are search terms. But NOW their algorithm will decide if you are large enough of a brand to warrant a reasonable way to be found in their galaxy.
All of the social media sites > amazon/ebay/twitter/facebook… are that big to where if they put a link up, it autmatically ranks ahead of any EMD domain if it has the same exact key words in a posting.
Not sure why no one is “worrying” about the Brands usurping small business even though the bulk of their content is relentless gibber jabber and my keywords contained within their jibber jabber with a zillion links ranks #1 now that Google will index it to keep them their.
Sorry for the length but I feel better. 🙂
In my experience it has been degraded from a 22% shot of a default #1 ranking to around a 17% one. The downgrade was more pronounced than I’d expected, though was mentioned by Matt Cutts fairly plainly in one of his YouTube Webmaster channel releases. I’m finding exact matches & near exact matches to still be good for a Top 14 or even Top 8 within 2 weeks of launch; all things being equal – which they never are obviously 🙂
Thank you. Good to hear that you’re feeling better after your comment
I think there’s a lot of paranoia when it comes to how a search engine might be biased, and that there are a lot of people who actively search out that bias regardless of whether it exists or not.
When I read an article that criticises Google and their “preference” for brands that doesn’t provide other possible reasons why the search behavior they are seeing might be happening, I’m skeptical. I’m a big believer in critical thinking, and I want to see other reasons explored as well.
For instance, I see a lot of posts about a preference for brands, but those never discuss how it’s about more than brands – that the behavior affects named entities – specific people, places, and things, including brands. It’s not a business based decision, but rather an algorithmic one. One that small business can be aware of, and use in addition to big businesses.
Whether or not a business uses exact match domains doesn’t tell us whether it’s a big brand or a smaller business, and the potential benefits of exact match domains don’t favor one type of site over another. A big brand can just as easily use an EMD as a small business.
A small business is going to have a number of disadvantages when it comes to business, but also has a number of advantages, such as the ability to make decisions and changes quickly, and to focus upon niches that are too small and too expensive for big business to explore.
From the patent, it appears that exact match domains shouldn’t work as well for highly competitive terms, but will continue to do so for much less competive terms, which still may receive a fair amount of search volume. That’s something that small business could, and should take advantage of.
I’m not sure that we have enough data to come up with the specificity that you mention when it comes to something like how much weight a specific ranking signal might have. When it comes to exact match domains, we don’t even know if its the domain name that gives it value, or the links pointing to it from other sites.
All things are never equal, and many of those things are things we know nothing about.
Hi Bill, although the weight of the patent leans heavily on the use of hyphens in a serch term, don’t you think this is already in the algo’s? After all, there was nothing to stop Google from implementing this, back in 2003, which I’m sure they did.
I have seen Matts video before, but still believe your average searcher will click on an EMD before any short, catchy attemt at branding.
I suppose, only time will tell.
The patent doesn’t really cover the use of hyphens within a query itself, but it does mention that it might be suspicious looking to the search engine if there are too many queries within a domain name.
I’m not really sure what your average searcher would click upon more when faced with search results that include exact match domain name URLs and URLs with brands in them. Google likely has access to that kind of information, but I definitely don’t. I would guess that most searchers will also consider what they see in page titles and snippets from pages as well, when deciding which results to actually visit though.
Such a good article. These are things I have been wondering and working on right now. Especially the hyphen thing, I have researched it and couldn’t come up with a good answer, I had read that google just considers hyphens as spaces, so didn’t know that if you had more than 1 or 2 that it could be flagged. Great article!
Thank you. Google does consider hyphens as if they were spaces, at least in file and folder names in URLs for pages and images.
I hadn’t seen anything before expressly from Google that said that they might look closer at domain names in them that have more than one hyphen, at least until I read this patent. I don’t think Google seeing more than one hyphen in a domain name might have the search engine label your site as spam, especially as it gets harder to come up with meaningful domain names that don’t have hyphens in them. But I’ll consider that when I need to choose another domain name.
I’ve seen this debate rage on and honestly it’s hard to tell if Google really cares at all about domain names, as long as they aren’t pure spam. Ive seen exact match dominate some SERPs, and fail hopelessly in others. While its inconclusive if it helps with rankings, Id say that it is a huge trust signal and greatly increases click through rates. Unless the site is competing in a vertical with very recognizable brands, I think that they would be seen as the most authoritative site by users. Just my two cents from what Ive seen in results.
I think we can tell that Google cares enough about exact match domains to have at least put out this patent, and come up with a fairly sophisticated method for trying to limit some of their impact.
I understand your point about how sites that use exact match domains are often focused upon those topics and can contain content that could be seen as authoritative, but there may be other sites that contain just as authoritative content where the owners weren’t first in line to get those domain names. Giving one site too much of a boost because of their domain name could potentially harm those other sites from being seen.
This debate about keywords in domain names is a matter of trying to draw lines that are as fair to every as they possibly can be. The patent tries to continue to give value to some exact match domains where the terms used aren’t too commercial, while limiting the value of keywords in a domain name where they might be.
It’s not a case of exact match domains trying to “trick” search engines.
It’s a case of the exact match domain closely defining what the content of the site is about (and therfore what it isn’t about).
StellarEnterprisesIncDOTcom could be about anything
Cheap-Laser-PrintersDOTcom is clearly about cheap laser printers and not much else.
I think that there is no point trying to pick a domain which search engines will favor. At the end of the day, if your business is fantastic, people will link to you because they actually like your website. Not because you are paying them or you are spamming their blog with your comments.
Keep it real, focus on your business and the links will come.
I think in the end, it really just depends on your business preference. If you are looking for search engine traffic, targeting a keyword might be a good bet. But if you are counting on other traffic including from offline referrals, you really should have your domain as the business name.
Google is slowly closing down all the ways a small business will get their website ranked – till now a lot of business rely on domain name and a bit of SEO to get their website ranked but if the value given to domain name is eliminated, it will be a big problem to most of small businesses.
If your brand name incorporated the keywords you would like to target, that would be the best workaround.
I don’t see why anyone would complain about this unless they are expecting an EMD to be a shortcut to the top. I read elsewhere this algo was aimed at ‘low quality’ domains, so it all comes back to the basic principle of creating a high quality site regardless of your URL. Over the years I’ve actually taken additional enjoyment in out ranking EMDs because it proves that a better site will always win. I’ve been doing that for a few years now so this isn’t new when talking about the power of a quality site!
The very first thing to think when creating a website is the domain name. This is the most fundamental thing that one must remember. One must understand that the power of domain name is that it gives your website a unique web address and other things associated with it. Perhaps it would be an advantage to choose a domain which is easy to spell, short, and memorable as suggested by this blog.
Toning down the dial for emd is great move by google. Most people noticed that they are giving too much weight for the keywords in domains. This is a thing that really bothers the people since this may cause burden in searching the things that they want. We hope that the newly patent exact match domain will be incorporated soon by google.
I’d definitely choose a domain that is “easy to spell, short, and memorable” in most cases. I suspect that many exact match domains for more commercial queries have had any extra value that might have come with the domain name dialed back.
Comments are closed.