Often times search engine optimization specialists will spend a great deal of effort trying to research and choose the right keywords for their optimization strategies.

After optimizing the keywords, code, content, and plethora of other on and off-site elements, the fun begins! The strategy is rolled out and ranking analysis takes over. Depending on how often/frequent the engines spider the site, chances are that the hard work has paid off and ranking improvements follow.

Being the ever vigilant search marketer, we know that ranking high does not always equate to higher volumes of traffic. So, turning to the analytics tool of choice, it’s evident that some of the newly acquired rankings are driving higher levels of traffic.

At this point, the cycle begins again. Research, implement, measure. Or, as we like to call it - wash, rinse, repeat.

After a few more wins (more high rankings, and increases in traffic) and several more wash, rinse, repeat cycles, you’ve made some gains, but the big breakthrough has yet to appear. What next?

By this point most SEO-ers have already scoured their analytics logs looking for diamonds in the rough. But, too often is this hunt for the golden keyword focused on the actual words themselves.

We’re not suggesting that this is a flawed tactic. On the contrary - it’s an excellent (and recommended) method of any sound organic search optimization strategy. But at the end of the day, this method is still an exercise in combing through mounds of data looking for a single piece of information.

One of the tools we use to aid in this process here at Azul 7 is a type of keyword analysis that raises the view a little higher. Rather than simply focusing on unique words or phrases, we’ll step out of the muck and analyze certain characteristics of those words. We’ll then compare those characteristics against a plethora of key performance metrics for the site as a whole.

The goal of this exercise is not necessarily find new words. Rather, it’s to analyze certain characteristics of words and how they map to more behavioral-oriented datasets. The words people use in their search queries can tell you a lot about what their intent is and what they are likely to do on your site. Because this search language is the literal manifestation of how people behave, it’s a wonderful method of helping to define a framework that can make your research more efficient and effective.

We recently performed an analysis for one of our clients. We’d achieved 200%+ increases in rankings across brand and non-branded terms, and though our target increases in organic search traffic have been achieved, we wanted to take things up a notch to the next level.

To begin, we downloaded the top 500 keywords used in Google searches from January 1 through September 1. Of course, this analysis will work on any size dataset, but the more data you have, the more likely you are to be unaffected by outlier data that can skew the numbers.

Next, instead of combing through 500 individual keywords/phrases, we analyzed the makeup of those phrases by breaking them down into different sets. The first set was just figuring out the distribution of words used in search engine referral traffic. We broke the full set down by groupings: how many referral phrases contained 1 keyword, 2 keywords, 3 keywords, etc. Our results indicated that of the full set, 36% of referral traffic used 2 keywords, and 45% of referral traffic used 3 keywords.

Keyword Analysis Tip #1: Do you ever find yourself debating whether to optimize for a 1 word, 2 word, or 3 word combination? Understanding the word count distribution can help narrow the focus!

Now that we had the distribution down, it was time to go a little deeper. We gained a tremendous amount of insight from the first step, but we needed more.

As part of our larger marketing strategy, we not only wanted to increase the number of leads this client was receiving from their site, but we also wanted to increase the quality of those leads. Thus, since we had established KPI’s as a method of measuring overall quality of site traffic, the next step was to map that distribution against those KPI’s.

Initially we discovered that 45% of organic referral traffic came from 3 word combinations. When we began mapping the initial distribution against visits from that source, we anticipated that 3 word combinations would likely make up a similarly high percentage of visits. However, we discovered that while 3 word combos were responsible for 23% of referral visits, 2 word combos were actually responsible for 54% of visits.

Keyword Analysis Tip #2: Mapping keyword distribution against at least one KPI will often reveal very interesting user tendencies.

This just wouldn’t do. We had 2 great pieces of data, but they were at odds with each other. It wasn’t quite clear enough at this stage for us to say with a high level of certainty that 3 word combos were definitely the way to go. From our first KPI comparison, it was clear that 2 word combos were actually generating more visits.

Since we had launched a completely new site with a considerably larger amount of content than the previous site, we’d identified early on that pageviews would be a key metric to observe as a measure of how much of the new content was being consumed. Our next step was to re-map the distribution against average pageviews to try and discern how the number of query terms impacted the amount of content viewed.

Ironically, we found that 3 word combos had a nearly 11% higher average pageview levels than 2 word combos. On small sites this difference in pageview levels usually doesn’t amount to much because the total pageviews aren’t that high to begin with. However, on our client’s site, the increased content had already demonstrated a significant increase in pageviews, so even a small difference of 11% could have a big impact.

At first glance we were somewhat relieved to see that 3 word combos seemed to be performing better. However, we were still only looking at average pageviews. In order to be conclusive, we needed to know how many actual pageviews had been registered.

When we computed the actual number of pageviews, we discovered that even though 3 word combos had average pageview levels 11% higher than 2 word combos, the actual number of pages viewed by 3 word combos was 50% less than their 2 word counterparts.

Keyword Analysis Tip #3: Expect the unexpected, but verify it with hard facts.

Throughout this exercise our goal was to gain additional insight into how we could make keyword research and ongoing optimization more effective. Finding exact words was key, but it’s also important to understand the context in which those words are being used. Armed with this data, the challenge of selecting and optimizing for certain keywords becomes less of a crap shoot. When you understand the context you can make better informed decisions about which keywords to focus on, and which ones to leave on the shelf.