By Paul Morris
I obtained a BIG data research paper from Mindshare/GroupM in partnership with HitWise pertaining to user behaviour in the SERPS recently and thought I’d share key learnings.
It was a paper focused on SEO yet it also provided useful CTR insight for PPC as well.
CTR Behaviour Research Methodology
For Google.co.uk over a statistically significant 12 week period ending 27/10/2012
Verticals studied were Travel (9.9 million searches studied), Retail (7.9 million searches), Automotive (4.8 million searches), Finance (3 million searches) and Insurance (1.3 million searches).
Key CTR Behaviour Graph: Search data by top level verticals
The paper went into much more detail per sector e.g. splitting out Finance into Mortgages, Loans, Savings, Cards and Banking however I’m not going down that route as this is a blog rather than a Research Laboratory.
…(My) Key CTR Takeouts:
- Some of the best SERP CTR Research I have seen so far (I spent a lot of time previously summarising other studies; you can find those in my Google SEO CTR Studies post)
- As can be seen from the graph above; whilst CTR generally reduces by position there are some anomalies (e.g. digging into the paper position 5 received more organic traffic than position 4 for ‘compare loans’ keyword and position 2 received more organic traffic than position 1 for ‘current accounts’ keyword) hence click behaviour is determined by the most relevant results and not just by position.
Slight digression: Google knows this they should be doing a better job for these anomalies and raising positioning for the lower ranked site that outperforms the site above as Google’s user base has voted and this parameter should play a larger role in Google’s algorithm.
- Linked with the above; the research project found higher brand recognition led to brands outperforming competitors in the SERPS.
- Interesting how much PPC traffic varies per vertical.
- Shows how important the top 3 positions are for CTR.
- The amount of traffic going to positions 10+ surprised me (higher than I expected).