Assessing Keyword Relevance
If you’ve been following my (occasionally interrupted) thoughts on keyword strategy, you’ll hopefully recall a couple core concepts: relevance and weighted popularity. You can read my full post on keyword metrics here.
The basic idea of weighted popularity is that you don’t really want to know how many times a given search term is used, you want to know how many times it’s used by people who are actually looking for what you have to offer. Taking a percentage to represent relevance, we can do a little math to estimate the number of targeted searches are available.
If, for example, we believe that 50% of the searchers using a given search term are looking for what we have to offer, and that there are 1000 searches per day using that search term, we can multiply 1000 by 50% to reach a weighted popularity of 500, or 500 targeted searches per day.
I’ve already discussed different ways to estimate the popularity of a search term, and I’m sure we’ll hit that subject again soon. What I want to discuss today is how to estimate the relevance of a search term. The easy way is to just take your best guess, but like a lot of things that are easy, this approach isn’t necessarily very rewarding.
Instead, I recommend a slightly more disciplined process which I will illustrate here.
One of the best tools that Wordtracker (and Keyword Discovery) offer is a comprehensive search tool that allows you to drill down on a specific search term, and find all of the search terms that include it. For example, if we look at “web hosting,” Wordtracker shows the following counts using their multisearch tool:
free web hosting 4231
web site hosting 1181
cheap web hosting 952
web hosting services 581
web hosting service 533
free web site hosting 359
business web hosting 328
adsl web hosting 258
web page hosting 240
affordable web hosting 192
professional web hosting 191
web design and hosting 184
There are actually hundreds more search terms listed by Wordtracker, but this is a blog, not science! Together, these search terms represent the different ways that searchers have refined their search for “web hosting.” This gives us a better idea of what people using that generic search term are actually after.
Taking these counts together, we have a total of 9,230 searches. It’s a lot easier to estimate whether these longer search terms are truly relevant. Let’s have our hypothetical web hosting firm review these terms and have them decide “yes, no, or maybe” on whether each term is relevant to their business.
For a “yes” answer, we will say that the term is 100% relevant, for “maybe,” 50%, and for “no,” 0%. Then we’ll multiply the count for each search term by those percentages to get a weighted popularity. The list below shows the search term, the percentage assigned, and the weighted popularity.
free web hosting, 0%, 0
web site hosting, 100%, 1181
cheap web hosting, 0%, 0
web hosting services, 100%, 581
web hosting service, 100%, 533
free web site hosting, 0%, 0
business web hosting, 100%, 328
adsl web hosting, 0%, 0
web page hosting, 50%, 120
affordable web hosting, 50%, 96
professional web hosting, 100%, 191
web design and hosting, 0%, 0
Adding these weighted popularity scores together, we get 3,030 targeted searches. Dividing 3,030 targeted searches by 9,230 total searches, we arrive at a value of 33% for relevance. Based on this exercise, we can (optimistically) estimate that 33% of the searches for “web hosting” would actually be relevant for our hypothetical hosting company, which is actually a pretty positive result for such a generic search term.
Before you dedicate a lot of resources to targeting very generic search terms, just because they seem to be extremely popular, it’s worth taking a few minutes to assess how relevant and targeted they really are.
Tomorrow I’m going to take a little side trip to look at an interesting article by search engine marketer Alan Perkins, which suggests that deceiving search engines (spamming) might actually qualify as deceptive advertising.