Many SEOs spend a lot of time trying to improve rankings for non-branded search terms, for all sorts of reasons. We do this too, but I’ve always kept a very close eye on branded search volume.
When we launched this blog in 2006 one of our primary aims was to improve our overall share of search. Another was to move the key brand metrics in a favourable way, not least because a visitor who adds ‘Econsultancy’ to a search term is 8-12 times more valuable than somebody who doesn’t include our brand in their query.
As such, branded search traffic is very important to us, but the horror show that is ‘Not Provided’ means that it is increasingly hard to track it. In fact, you will be appalled if you only look at your analytics data.
With this in mind, I thought I’d show you our numbers, and provide a workaround for you to try.
First up, let’s look at the rise of ‘Not Provided’. Unfortunately this has now grown to about 68% of all organic search referrals within two years.
Now let’s figure out what our branded vs unbranded search looks like.
If you use Google Analytics then aim for ‘Organic Search’ (in Traffic Sources > Search).
Next, click the ‘advanced’ filter to add your brand keywords. For starters, your company name, but it’s also worth adding your sub-brands, such as products and services.
You can choose to include or exclude these keywords, depending on which view you wish to see (branded or non-branded).
Having done this we can start to look at trends, and I’d recommend that you download a few .CSV files along the way, as you will need them later.
Here’s a look at our branded search traffic in the two years to the end of June. Note the savage decline in volume. We can thank ‘Not Provided’ for that…
It all looks very bleak, but the reality is somewhat different, and things aren’t as bad as they may appear. What we need to do is offset all of that ‘Not Provided’ traffic, to find some more accurate numbers.
To do this, you’re going to have to open up a spreadsheet. I’ll walk you through it…
The goal here is simply to remove the ‘Not Provided’ traffic from the equation, in order to try to find a much more realistic branded search volume.
The chart below shows our analytics data for the past two years, with ‘Not Provided’ traffic and branded search traffic split out.
An explanation of what’s going on in the columns…
Total Search Volume: The number of visitors referred by organic search, per month.
Not Provided Volume: The number of visitors who arrive without any search query data.
Not Provided % Of Referrals: The percentage of ‘Not Provided’ traffic.
Branded Search Volume: The total number of visitors (that we know about) who included the word ‘econsultancy’ in their search query.
Stated Branded Search %: The percentage of branded search traffic according to Google Analytics. However since it includes ‘Not Provided’ traffic, this number is pretty much meaningless.
And the two most important ones…
Actual % Branded Search: The percentage of branded search traffic when Not Provided traffic is removed (Branded Search Volume as a percentage of Total Search Volume minus Not Provided Volume). This is the number we want to uncover.
Probable Branded Search: Total Search Volume multiplied by Actual % Branded Search. A much more likely number than the 6,600 suggested for the month of June 2013 by Google Analytics.
A few observations…
- By my reckoning branded search may have actually fallen to just short of 8% of total traffic. I would have expected to see a slight decline, given the rise in overall search traffic (which has almost doubled in two years).
- In terms of overall volume, we’re probably averaging more than 20,000 branded search visits a month. Better than where we were at in the second half of 2011, when Not Provided was introduced.
- As Not Provided grows, the sample size reduces, which is a threat to accuracy. I still think there is enough data to make some reasonably accurate guesses, but when Not Provided accounts for more than 90% of traffic it could become more difficult to make sense of things.
Anyway, that’s how I do it, and I’m hoping that a statistics wizard doesn’t suggest that it’s all shades of wrong (but please do if my brain has gone off piste again).
Do you have any other ways of measuring branded search? Please leave a comment below if you have other suggestions.