This is a companion article to the video below on measuring SEO results with web analytics. It is meant for people who have already watched the video.
The first section is for people who want an easy-to-follow list of the steps I run through in the video. The second section is for people who want further explanation behind the steps as well as alternative approaches.
Step 1. Select Before and After Periods
- Within your Google Analytics account, go to the date range section.
- Select your “after” time period.
- Check “compare to past.”
- Select your “before” period.
Step 2. Select Non-Paid Keywords
- Go to Traffic Sources and select the Keyword report.
- Click the non-paid link.
Step 3. Filter Keywords
- Click Advanced Filter.
- Add a filter “containing” your keyword theme root (you may need a second contain filter).
- Add another filter “excluding” your brand name root (you may need additional exclude filters).
- Scroll to the top of the page and find the before-and-after results for visits.
- Enter the visit counts in a spreadsheet, along with the keyword theme and the “root” that you used.
Step 4. Measure Quality
- Find the “after” period numbers for Avg. Time on Site and Bounce Rate for your selected keyword theme.
- Open Google Analytics in a new tab and go to the dashboard view.
- Find the “after” period numbers for Avg. Time on Site and Bounce Rate on the dashboard view (will be for total site traffic).
- Enter the quality metrics in your spreadsheet, next to the visit counts.
Step 5. Repeat Steps 3 and 4 for Each Keyword Theme
Note that, for Step 4, you don’t need to repeat the “dashboard” steps since you already have average site metrics.
Bonus Step: Package It
- Add “percent difference” calculations to your spreadsheet for 1) before vs. after visits, 2) site avg. vs. keyword bounce rate, and 3) site avg. vs. keyword avg. time on site.
- Create a graph to display the metrics.
Additional Tips, Explanations & Alternatives
On Picking Before and After Periods …
Choose time periods that 1) give you as much data as possible and 2) are as apples to apples as possible. Most businesses have some sort of seasonality; to remove this as a factor, pick the same time period from a prior year when possible.
On Removing Paid Search Traffic …
If you are doing a lot with PPC campaigns, selecting “non-paid” search traffic is critical if you want to do an accurate analysis of your SEO. But Google Analytics does not automatically separate paid and non-paid search traffic. If you’re running AdWords, you need to link your AdWords and Analytics accounts for your PPC traffic to show up under “paid.” If you’re running paid search campaigns with Yahoo or Bing, you need to add tags to your URLs.
On Using Keyword-Level Analysis and Excluding Brand Searches …
If you’re like most companies, you don’t do SEO to get more traffic on brand searches; you already rank well on those keywords. Rather you optimized your site around non-brand keywords. When doing this type of analysis, then, you need to be diligent about removing brand search traffic. Brand traffic is often much higher volume and variable, especially if you are doing offline marketing or getting press. If you include it, your picture of your SEO will be distorted.
In the video I take the approach of drilling down to the individual keyword (theme) level to do your analysis. This is almost always an effective way to evaluate your SEO. But what if you want to see metrics for your target keywords as a whole? You can sort of get at this by totaling up your individual keyword metrics. But there will be a lot of overlap among your individual keyword counts. A potentially more accurate approach is to go the other direction and use “exclude” filters to pull out as much non-SEO-related traffic as possible. If your site’s keyword traffic sources are limited, this is probably do-able by trial-and-error. Remove your brand keyword traffic, see what other non-SEO keywords are sending traffic, and then remove them. But if your site gets traffic from a large number of keywords, you’re better off totaling up your individual keyword numbers.
Note that, even when I filter at the individual keyword level, I still add exclude filters for brand keywords. That’s because there is often crossover between target keyword searches and brand searches (e.g. “affinity lab shared office space”). We want to exclude these searches from our SEO analysis because most likely they already ranked well on these searches.
On the Downsides of Using Rankings to Measure SEO …
- Rankings fluctuate … both over time and among users. Your geographic location, past searching and clicking behavior, and other factors increasingly influence the search results you see, meaning that your site might appear #1 for you on a given keyword but #20 for someone else (more on this here). And tomorrow it might appear #5 and #40.
- Many if not most searches are on “tail” search terms. When you use rankings as a measure, you’re relying on a pre-determined list of keywords. Even if your list is extensive, it inevitably misses many of the actual words people are searching on. For every person searching for “soccer cleats,” there’s someone searching an obscure variant like “soccer cleats for wide feet and outdoor turf fields”. After some SEO, your site might still rank poorly on “soccer cleats,” but could start getting lots of traffic on the “soccer cleats” variations that you never could have thought of in advance, and that traffic will tend to be higher quality. Rankings will miss this as a success; Google Analytics will capture it.
- Rankings don’t equal traffic — or quality traffic. People usually track rankings as a proxy for traffic. But with Google Analytics, there’s no need for a proxy because you can track traffic by keyword (including all variations of a given keyword if you use it the right way). And if you can see which traffic is high quality and which is low quality, by keyword. Sometimes a #1 ranking drives surprisingly little quality traffic. We’d much rather know that a keyword plus its variants went from driving 0 to 5 leads per month than know that it went from #15 to #5 on Google.
This is not to say rankings are a useless measure. They can help to identify opportunities for SEO improvement, and we tend check them once a quarter or so as an additional gauge. But when it comes to measuring progress on a regular basis, they’re much less meaningful than web analytics data.
On Choosing the Right Filter Words …
With most of the steps above, you are narrowing down your traffic. But given the way people search, if you filter on exact keyword phrases, your results will be too narrow. So for filters you need to broaden your scope a bit.
Take your target keyword list and create two additional columns: “theme” and “root”. Scan your keyword list and find the common threads among them and enter those under the “theme” column. Then think about other relevant variations of that “theme” word and about what the smallest possible root of that word can be to pick up those variations without picking up unrelated words. Add that under the “root” column. Getting the right root often takes some trial and error.
On Measuring Quality …
For many businesses, the best way to measure the quality of your search (and other) traffic in Google Analytics is by using conversion goals. If you have conversions goals and place and there is enough conversion data to be meaningful, you can just select “Visits with Conversions” under Advanced Segments (upper right) once you have your filtered list in place. In our video I focused on Bounce Rate and Average Time on Site because so few businesses have conversion goals in place, and because even when they do the data is often to thin to be useful in this type of segmented analysis. Even when you are using conversion data, additional metrics like Bounce Rate and ATOS can be useful supplements to your analysis.
In the video, I advocate comparing “target keyword traffic” vs. “average site traffic” (rather than “before” vs. “after”) when you’re using things like Bounce Rate and ATOS. If you’re using conversion goals, however, “before” vs. “after” is the way to go.
On Alternative Approaches to Measuring SEO …
The approach I use in the video is not the only way to measure SEO in Google Analytics. But it is I think the most straightforward, especially for people who are not advanced GA users. If you are using this approach often and/or have a very long list of target keywords, Custom Reports may be a more efficient approach. We’ll cover this in a future video.
Want to get more tips like this? Subscribe to The Keyword, Marketade’s free email newsletter.