Data-driven analysis SEO tools have become very popular over the last few years.
These are tools that analyse the top-ranking pages for a given keyword and offer suggestions about what may be needed to outrank them. Most of the recommendations are on-page, so changes can be made immediately for some potentially ‘quick-wins’.
I wanted to find out if these tools really work and the kind of results that can be expected from using them.
I decided to test out SurferSEO and POP by implementing their suggestions on 4 pages (2 each) on a small website I own.
The website was already ranking in the Google SERPS and I chose keywords/pages that were currently ranking outside the top 10 but inside the top 50.
The keywords all had a lowish search volume and had a low-medium ranking difficulty (although they were also in a competitive YMYL niche).
NOTE: The other big player in this field is Cora, but I didn’t test this as it is only available for local download on Windows and IOS and I am currently using (and loving) a chromebook. Surfer and POP are online tools.
The table below shows the initial ranking of each of the keywords and the tool I used to optimise the content.
|Keyword||Initial Rank||Final Rank||Tool Used|
I ran each of the keywords through the relevant tool and implemented all the recommended actions. I then waited a week or so to see what happened…
Looking at the results, Surfer SEO had more of an impact than POP.
The two keywords that I ran through POP seemed to have very little effect, however the two keywords that went through Surfer SEO had a much bigger impact on their search rankings, jumping higher up the SERPS.
Now, this is only a small study and, although I tried to keep them fairly even, there are a number of other variables outside of my control that could have affected the results – for example, the competition for the keywords that POP is targeting could have ramped up their SEO.
Although I use Surfer SEO regularly and love it (it has a much slicker interface than POP as well) I don’t think that POP is as bad as this study would indicate. I’ve used it before and it usually gives much better results – I was quite surprised with the outcome.
So, I think in fairness, I need to do another study with a much larger dataset to eradicate outliers – it’s on the todo list.