One of the first and most obvious types of data we want a search term is just how popular it is with searchers. This is important for all search engine marketers, whether they are working on SEO or PPC campaigns.
Unfortunately, when it comes to keyword popularity, the data that is available can be hard to trust. Today, I’ll walk through the main search term databases that are available today, and discuss the specific pros and cons of each one.
All search term databases are affected by a certain amount of skew. The main problem is rank checking applications like WebPosition, which submit automated queries to multiple search engines for large lists of search terms. Every time someone checks their rankings (even by searching manually), it inflates the count for that search term at that search engine… and sometimes at others as well.
The first tool most folks use is Overture’s search term suggestion tool (Google it, I’d link but they’ve changed the URL more than once). It’s free, so it’s popular. Unfortunately, “free” is the best thing you can say about it if you’re planning to use it as a keyword research tool for SEO. Why is it so hard to trust for SEO?
Skew in the Overture tool: Pretty much all rank checking applications will hit the major search engines, and because Overture’s ads are carried on two of them (MSN and Yahoo), every rank check counts as 2 searches to Overture. Other applications that check PPC bids may also run queries on Overture.com, further inflating some numbers. Overture also has partners like Claria serving up search results on pop-up and pop-under ads, search results displayed on expired domains, etc. These all count as “searches” to Overture, even though they don’t really represent a person conducting a web search.
Search terms in the Overture tool: Because they are purely a pay-per-click service, Overture does some things to search terms that aren’t really helpful if you’re using the tool for SEO. Singular and plural versions of a search term will be combined into one, which is important because many times only one (singular or plural) will be really relevant for a web site. Some long search terms may actually be combined – such as “honda car red” and “red honda car.” The search term displayed on Overture’s tool may actually represent a composite of many search terms. This matters little to PPC ads, but it’s a deal-breaker for SEO.
Most professional search engine marketers use the Wordtracker keyword database these days. This database is compiled out of the prior 60 days worth of searches on the Dogpile and Metacrawler metasearch engines, which reduces a lot of the skewing caused by rank checking, since most rank checkers don’t bother with metasearch engines. (Metasearch engines combine the results from several search engines, so their rankings don’t really represent anything that’s actionable for a search engine marketer).
Because Wordtracker’s data is relatively “clean” when compared to the others, we use it in a lot of ways, from helping us identify search terms, to assessing the relevance of search terms. Wordtracker also provides competitive data from many search engines, including pay-per-click bid prices from Overture.
A relative newcomer is the Keyword Discovery database, which is operated by Trellian, the parent company of the Priority Submit paid inclusion service. Trellian’s database is unique in that they have over a year’s worth of data. This allows marketers to search for keywords that are seasonal in nature – if you need to know what popular search terms were for Valentine’s day gifts, but it’s November, you won’t get that data anywhere else.
Keyword Discovery’s database is very large, containing over 11 billion searches, from… well, “from where” is something we don’t know and they can’t disclose. As a result, we don’t really have as much trust in the search counts provided by this database, or even the data on seasonal trends, since we have no idea how many search terms are in the database for a given month. Since we don’t know the size of the database for a particular month, it’s very hard to project results of a search engine marketing campaign – there’s simply no way to get from their count to an estimate of how many searches will take place in the next month.
Aside from having seasonal search terms (even with unreliable counts, it’s better than nothing!), the major advantage Keyword Discovery has over Wordtracker is the tools that they’ve developed to help marketers carry out research. These include an excellent “related terms” tool, and a URL analyzer that will extract search terms from a web page and present the list along with the actual counts from the database.
MSN’s bCentral also has a keyword research tool offered with their monthly subscription service. Unfortunately, users must jump through so many hoops to get to it, and the data provided is so shallow, that I can’t recommend it at this point. I will keep an eye on it, though, and let you know if it gets better.
My next post will cover the recent elections at SEMPO, the Search Engine Marketing Professional Organization, then we’ll move back into keyword strategy with a look at ways to assess the relevance of search terms.