The DomainRoundTable Conference hosted Matt Cutts of Google and a panel of SEOs, to discuss search marketing and domaining. One topic of conversation afterwards was the value of keyword domains. Is a domain “better for SEO” if it contains exact match keywords?

Matt said yes, keywords in the URL can help rankings, but you don’t need that in order to rank. I would like to expand on that, as I understand it. Since I blogged about Matt’s comments, I have seen several SEO discussions online that I think demonstrate a lack of understanding of what matters with domain names (as relates to SEO) including Matt’s comments. Please comment if you have more to add, or can help me better understand your own knowledge or experience.

Consider the search user. If the searcher asks Google “where can I find books for sale online” by entering the Google search query “buying books online“, Google will serve up a set of results it believes will be useful to that searcher, and which will engage that searcher. Google considers various factors in making that results set, but when it displays that results set to the user as a SERP, the destination URL is displayed prominantly as the name of the listing. That name is the only authoritative identity guaranteed to be present in the SERP for that particular web page.

When the user scans the result set, that user has an expectation, which we assume is aligned with the query “buying books online”. When Google’s customer sees the URL contains the keywords they are searching, that keywordy URL reinforces their hope that the results set provides that which they seek. But it is more complicated than that. The substance of that noticed match must also convey a sense of safety and possibly authority, or opportunity.

There are various facets of user action motivation, but for now let’s just accept that Google anticipates that Google’s customer will trust “OnlineBookStore.com” more than they will trust “buy-books-here-discount-reduced.info” for the query “buying books online”, all other things being equal.

And so Google will give an advantage to OnLineBookStore.com in this case. Keywords in the domain, yes, but more importantly, authority and trustworthiness expressed in the domain name. The keywords added authority (via relevance).

What if the domain is books.com? That is a very expensive domain. Simply because it is expensive, I believe, Google gives it more trust (all other factors being euqal.. such as content and reputation). So in theory, everything else being equal, is books.com is better for seo than onlinebooks.com, for the search “buying books online”?

I don’t think so. Due to the keyword match of both “online” and “books”, I would give the edge to OnlineBooks.com. I don’t think the added value Google might give Books.com is adequate to overcome the value of the easy to remember and well matched OnlineBooks.com URL. That is my opinion, limited to this example.

What about “BuyingBooksOnline.com”, you ask? That’s an exact match to the query. Yes, it is, and that is good. But…. Google is focused on Google’s customer. Books.com is easy to remember, easy to recognize, and authoritative. The user will appreciate it differently than BuyingBooksOnline.com. It is up to Google to decide whether or not users will appreciate BuyingBooksOnline.com as a destination for the query “buying books online”. I would assume there is benefit from the exact match, and benefit from the user experience associated with soley that domain name displayed in the SERP, and that the combination helps Google rank the results set. I would place OnlineBooks.com ahead of BuyingBooksOnline.com, all other factors being equal for this one query, because I think it is a but more accepted by the user… I think they would view BuyingBooksOnline.com as more of an information site than a book store.

So what about OnlineBookstore.com, you ask? Again, all other factors being equal (trust and reputation, backlinks and content, etc), I have to consider the cultural aspects of the Google customer. In what context does the Google customer refer to “bookstore” as compared to the context for the use of “books” in our language? College students refer to “the book store” quite frequently. Gift givers look for books more than they look for book stores. OnlineBooks may refer to PDF files you can download instead of books you hold in our hand. And store is often a synonym for shopping cart or ecommerce web site. That’s right… I am suggesing that Google incorporates cultural sensitivities into the ranking algorithms. if for no other reason than because the corpus of information Google uses to determine relevance comes out of our culture: the index of documents studied by Google. We, like Google, have to consider the context of our question “buying books online” when sorting through available web pages and compiling a set of search results.

Based on my perspective, I’d give the edge to OnlineBookStore.com over OnlineBooks.com, except in the context of college life I’d definietly go with OnlineBookstore.com. That’s an opinion… I don’t have Google’s resources. But I think it demonstrates my thought process, and maybe you can see how it would be applied to other search queries where you and I may have significant insight into the vernacular of our customers. Google wants the search results to be good for the user experience, and if the URL adds to the user experience (or detracts), Google wants to consider that in the ranking/scoring algorithm.

But does Google do that? Theoretical discussion are great for academics, but does this really matter? Does Google count it? Yes, Matt Cutts tells us that when the domain adds to the user experience, there is an advantage (all other factors being equal). And he said so in the context of a question addressing whether or not keyword domains are better for SEO.

Matt won’t tell us if Google discriminates between OnlineBooks.com and OnlineBookstore.com, but our example of assuming “everything else being equal” is a tough constraint anyway for our practical world. Small differences in back link quality or quantity or content quality are probably more important than the difference between OnlineBooks.com and OnlineBookstore.com. And that’s an important point!

Maps.com is a great domain name. But MapWorld.com may be better, because it contains the second keyword “world” which matches queries that contain “world”. It also suggests a content theme “world” which is semantically associated with many search keywords commonly paired with the word “map” (such as “maps of the world” and “maps of north america” because of the semantic association of “North America” with “world”). Think about that search index.. when Google tries to guess searcher intent for a query, it looks at word associations based on the indexed documents and their relationships.

Again… there is so much more being considered by Google for ranking web pages, but the domain name “MapWorld.com” contributed significantly to user experience and semantic meaning in the context associated with a large number of search queries. That adds value. For a global map business do you need maps.com or can you do it just as well on MapWorld.com for a much lower initial price for a domain?

I know this is complicated and somewhat subjective. That’s part of the point. It’s not all about the domain name, but domain names that carry meaning for the searcher do have more value, within a specific context. Think user experience… user experience… user experience, all other factors being equal, and then make sure all other factors are not equal, in order to compete.

ISPs are placing the security and privacy of their customers squarely in the hands of a third-party ad company

Dan Kaminsky is always fun (and scary) to watch, and he is describing how some ISPs are utilizing a 3rd party DNS service in an attempt to monetize user activity. According to Dan, that third party is exploitable (XSS vulnerability), and this outsourcing action is putting ISP subscribers at risk. Via Wahington Post.

This is live blogging from matt’s session. It will update as I enter information.

Matt said: about his role on the web spam team, he defined spam as sites that “rank higher than they deserve”. Go figure that one out.

Matt asked how many people were into domaining because it was like a garage sale, where you find a rare book worth a lot on sale for only 50 cents. He also asked how many were in it for the money, and then how many had a life long commitment to creating new content and publishing content of value to users on the web…. Can you see where Google is going with this? Sure you can.

He highlighted GMHS.com which is a for sale domain. User won’t be happy if they typed in the name of a high school and got this. Not too happy a user.

Earthday.org parked page… says it is relevant. Lots of user stuff. Complete new user might be happy landing on it. A savvy user will not be as happy with it.. they will wirte in and say… Matt suggested hiring a blogger to be the EarthDay Blogger for 10% of the eventual value… cherry pick your top 10,20 domains and give a blogger some equity to write content.

Ajaxian.com neat site about AJAX. Take gmhs.com and get somebody to develop it.. that’s the high end of content and value add, because not everyone is providing that. For the valuable domains, that is what Matt would do.

Q for Matt: standard dupe content question. Matt says he can handle that. Litmus test is “were was the first place this content debuted (was viewed)”. Gigablast is like 2 guys and can’t do that, but Google can. Google filters out dupe content that is not as useful as the original. What abut shuffling content, dictionaries.. trying to evade detection, as Matt says. He says it is easier to find someone to generate that content for you.

Q: on DMCA process from Ron Jackson, do you complain to Google or the host? Matt says google.com/dmca.html to describe that process. There is a process for counter-notify and dispute, and if that happens Google stops and leave the debate for the involved parties to handle.

Q: from a lawyer… an admittedly frustrated lawyer, not having great success because people just switch web hosts when challenged. Matt says Google “doesn’t ant to play police”. The lawyer says Federal copyright registration is a prerequisite to DMCA, and not easy to get a copyright on a web page. Matt suggests that after you’ve been scraped a few times…people look for ways to embed links in the article to take advantage of the scraping… “I get a lot of links”…”I’m guaranteed to have more page rank than they do”… he personally says “oh well, that’s links that go to my website”.

Q; on tld’s and their impact on ranking. Matt says early literature shows G didn’t care about what TLD was using.. just # links and how reputable those links were. He says except fro some corner cases, it doesn’t matter, and he says most people will never fit those corner cases.
Note: Matt says the new york times is more reputable than your college friend (he was addressing link value). Think about that.

Matt: “you never want your users to be angry” , Matt remembers his mother in law with a huge infection of scumware, and how much Matt spends the first day of a visit cleaning up her computer. Some people don’t want their ads showing on parked pages. Matt says Google helps show people how the domain channel can work as a profitable advertising channel.

Q; about how long it takes for a new site to monetize. Taking longer now than it used to. Matt says people think a page gets a little pagerank just because it is a page, which is a misconception. Page rank is peanut butter… you’re spreading it around, it gets thin. You need more links (more peanut butter?). Think about marketing aspect.. catchy angle that attracts people’s attention, and then spread that around your network. Q: Gestation period has gotten much longer…. Matt says it can take time for pages and trusted pages to develop.

Matt showed off searchmash.com. Will we see some of these features on Google? Entirely possible. Notes the integration of DomainTools for whois as of yesterday. “please don’t scrape this”…. Google has built in a “fair amount of checking” so too frequent queries will cause it to block you. “We like this idea of trying out experiments”. He searched “aa 127″ and got American Airlines flight status for flight 127.

Matt says if a domain changes hands, Google resets the links vale to zero/near zero. [Update: Matt apparently said this about expired domains in 2007. I can’t be sure of exactly what was said here, but these were contemporaneous notes so perhaps we will have to wait for the recorded sessions to be sure].

Domain names are the primary way of mapping where domains are on the web and Matt expects that to continue. Domain names are important and inseparable going forward.

Generic domains that users are likely to remember, will indeed carry more weight than others. There is a real value to those FuneralHomes.com for example. Google does give keywords in the URL a certain amount of weight, but you don’t need it in order to rank.

“We have a deal with GoDaddy that if you sign on with GoDaddy you’re automatically registered with Webmaster Tools”.

Q: Parked Domains: ” We try to detect parked domains, and once they leave their parked status, we let them in relatively quickly”

Q: If a domain says it is for sale, does that harm it’s chances in Google? Matt: Our litmus test is not whether or not it’s for sale, but if their’s good ocntent on it and it’s helpful to users.

Q: if you stub your toe [violate google guidelines] on on domain of thousands, do all of their domains suffer? Matt says no.. just because one domain is doing something bad…. BUT, it does increase the odds of google scrutinizing the other domains. Says google knows how to find other owned domains via common templates etc. If just doing everyday stuff, one domain in trouble doesn’t hurt other domains.

Q: Breakup page of more than 100 links… people complain about it.

Q: Ip cloaking to block abusive users. Matt says be careful.. ok to block scrapers etc but Google runs spot checks from different IPs… matt will go to his old school account to see what the page looks like. If user and Googlebot see same thing, should be ok. Matt cares about cloaking Google, not other users. BUT be careful not to get it wrong.

Q: Geo IP cloaking question… Matt says ” different MD5 sum means high risk category” ;-) Dont treat Googlebot like it was it’s own unique country (Googlestan), getting Googlestan content. We crawl from California… if you cloak it, be very careful to say what you are doing “it looks like you are outside of Colorado..so we’re serving you outside of colorado content…”

Q: on use of nofollow. Directory owner, asking if nofollow helps or hurts. Matt says nofollow is a “very simple thing”. Nofollow link doesn’t flow pagerank, doesn’t flow anchor text. Link level to say “I trust this link but I don’t trust this link”. You don’t want to flow page rank through them if you don’t trust them. Real business 3-4% of your links will be stale, don’t worry don’t need nofollow. If check them at some point, willing to vouch for them, at some point checked them for quality, then don’t need to worry about nofollow. If just a domain directory, use no follow.. it is a matter of how much due diligence you put in.

Q: Webmaster asked about DiamondsDirect.com and why it and other sites don’t appear in Google. Matt looked at it, said the site was good, most users would lik eit, but the feed data was dirty (some control characters showing up) and appeared at many places.. probably more unique content.