There are hundreds of keyword research tools out there. Some show you how popular a keyword is, or even its monthly search volume. Others focus on the competitiveness of a keyword. There are many that will “grade” a page on your site for a certain phrase, or alternatively determine what phrase a page is focused on. There is no “right” way to do keyword research, and no “right” set of tools. Some tools are better than others, just as some methods are better.
Hundreds (thousands?) of articles and blogs have been written on how to properly research and choose keywords. Most of the time one of the best resources available to you is left out of the mix: your analytics data. This article will focus only on basic tips for using your Google Analytics data for keyword research. A big disclaimer is needed: Analytics should not be your only stop for researching keywords. It does not indicate the popularity or competitiveness of a keyword phrase. However, it can provide valuable insight into how visitors are finding your site and what they do once they are there, based on keyword data.
First things first. You need to define a time period with enough data to draw conclusions. This will depend on how trafficked your site is. I recommend no fewer than several thousand visitors; I usually aim for at least 10,000. You can use these tips based off keyword referrals from all search engines, although narrowing it down to only 1 is best. Do that by navigating to Traffic Sources -> Search Engines and then clicking on your search engine of choice (Google). You will now see all of the traffic coming from Google over your selected time period.
Keyword Grouping
The problem I find with this list of keyword referrals is that similar keywords are not grouped together. In the case of our client, which we'll call Widget Arts and Crafts, they get visitors searching for “widgets”, “widget arts”, “widget crafts”, and “widgetarts.com”. If we want to see how many visitors are finding them based on their brand name, we need to group all of these together. Similarly, most websites have natural sections or divisions within them. For ecommece sites, this breaks down easily by the type of products. I may want to see how many visitors find this site with searches related to shirts, or pillowcases. The easiest way to find this information is by putting in a simple keyword filter at the bottom of the list of keywords. When using keyword filters, try to be as generic as possible. Don’t narrow it down to “t shirts”, but only “shirts” to get an idea of how many visitors find the site looking for shirts.
This is unfiltered traffic, a “control” sample of sorts
Next traffic is filtered for “widget”. As expected, brand related visitors are typically return customers or people who know what the site is. They should typically be more targeted, have higher pages/visit, Avg. time on site, a lower bounce rate, and a higher conversion rate.
Filtered for "widget"
Now filtered for “cutout” - a word that indicates a particular line of widgets they sell. I can see that these visitors are also very qualified and finding what they’re looking for. We expect their conversion rate to be above the site average also.
Filtered for "cutout"
Goals for "cutout" keywords
Here is an example of an under performing keyword group. This time, we filtered for “craft”. (We also excluded the word “widget" to filter out the brand searches. You can input two keywords using an advanced filter.)
Filtered for "craft"
Here is the conversion rate based on those filtered terms. It is low, as expected.
Goals for "craft" keywords
So how do we use this information? This data is infinitely more useful than just looking at a specific phrase. “blue cutout widgets” may only bring in 200 visitors, but other cutout terms bring in a significant portion of traffic when looked at as a whole. We might take that information and change some of the keywords we target to less generic craft terms and focus more on increasing our rankings for cutout phrases. If “stencil” phrases are getting a lot of traffic but not selling well, we can encourage the customer to run a special on them. A “buy 2 get 1 free” special, or a graphic on the home page might have a big impact on stencil conversions.
High Converting Keywords
This is one of my favorite reports to analyze. It assumes you have a goal setup. Some non ecommerce websites won’t have any goals in analytics, but instead of conversions you can use time on site or pages/visit. Also, this report does not work as well for smaller sites that don’t make a lot of sales. For example, Widget Arts & Crafts only has 11 keywords that, since Jan. 1, brought in 100 visitors and made at least 1 sale. If your site is bringing in 100,000 visitors you will find this report extremely useful to find out your most important keyword terms. Maybe there is something you didn’t know about before.
On the same page as before, clear your keyword filter and click “Advanced filter”. Filter for “Visits” and “Goals” (whatever you named your conversion goal, in our case we usually name it “Checkout”). As I said, if you don’t have goals setup you can use a different metric and get similar results. Put in values that are significant enough to narrow down your results to a manageable list. For the site with 100,000 visitors a month, you may not care about keywords that only make 1 or 2 sales a month, but only your big terms that make 20 sales a month. (This isn’t to say those long-tail phrases are unimportant, it is just difficult to focus on all individual long tail phrases rather than grouping them together.) In that case, try filtering for 1,000 visitors and a 2% conversion rate. The smaller site, like Colortime, might want to filter only for 50 visits a month and any conversion rate other than zero.
This report is a goldmine. I try narrowing it down to 25 or 50 keyword phrases that are big money. Particularly for high traffic sites, this can tell you what some of your main keywords are. You may be surprised to find out that even though you’ve been targeting “blue widgets” and have it ranking consistently in the middle of page 1, “blue cutout widgets” is actually bringing in more traffic and making more sales.
So how do we use this information? Not only can this report give you ideas for valuable keyword phrases that you have not been targeting, but it can be a great spreadsheet to give your client showing that the phrases you are targeting work. If the top 20 keywords in this list are ones that you have been targeting, you’re probably doing a good job with them. This list is better than just sorting by conversion rate, because it helps you filter out those 50% conversion rate long-tail searches that only bring in 2 visits but make a sale.
Maybe the phrase you’ve been working on for months and just celebrated because you moved to the first page doesn’t make the cut. Is it not bringing in the traffic? Or is it not making the sales?
You can also export this to a spreadsheet in order to look at it side by side with ranking data. It’s great to show that you moved the website from ranking #3 to #1 for a high volume phrase. What the client really wants to know is how many more sales this brings them. Last month at #3 it brought you 1000 visitors and 20 sales, this month at #1 it might bring in 5000 visitors and 90 sales. That’s what is important to the client and what search rankings are all about.
Keyword research is a vital part of SEO. Looking at search volume, rankings, or competitiveness is a great first step. Analytics data shows real results that are the culmination of all of that information. Don’t neglect to include it in your research. This was only 2 simple methods for looking at your keywords in a new light. There are many other approaches to using Analytics for keyword research. I always encourage exploring new ideas and hope I got your wheels turning. Look forward to more Analytics tips in the future!