I love jumbling bits and pieces of data around to see if I can distil any meaningful insights from them.
I have absolutely no training in data analysis, so my amateurish efforts at detailed analyses are probably laughable in the eyes of seasoned data analysts. But I enjoy doing these things, and I rarely care what others think anyway.
For a few months now I’ve had filtered Google Analytics profiles for a handful of customer websites aimed specifically at judging and subsequently mitigating the impact of Google’s rollout of SSL search.
One of the filters active in all these profiles is aimed at extracting the Google rank of a keyword when a user clicks on it and arrives on the website.
Now with several months’ worth of keyword ranking data, I figured I could try to analyse this data and see if it can tell us anything interesting about rankings, visits, and click distributions.
So I extracted all the keywords from the relevant Google Analytics profiles in to Excel sheets and started juggling the data around a bit.
Specifically, I matched up the number of visits on a given keyword to the rank of that keyword (as passed on to Google Analytics in the referral string). The picture that emerges should not be a surprise to anyone:
This graph shows that the #1 rank is still the champion of search engine rankings. #1 ranked keywords generate the bulk of the visits to a website, as is evident from the % graph as well:
On average across my client sites, we see that first ranked keywords on Google account for almost 40% of all Google visits to the sites, while only making up around 18% of the total number of keywords that are used to find these sites. There are some fluctuations between different sites, but none stray far from these average figures.
From position 2 downwards we see this long tail of traffic even out rapidly:
|
Rank |
% Kws |
% Visits |
|
1 |
17.83% |
38.71% |
|
2-5 |
35.35% |
28.97% |
|
6-10 |
26.34% |
18.04% |
|
11-20 |
14.11% |
10.77% |
|
21-50 |
4.88% |
2.73% |
|
51-100 |
1.03% |
0.53% |
|
101-200 |
0.37% |
0.19% |
|
201+ |
0.11% |
0.06% |
More than half the number of keywords that send traffic to these sites are ranked in the top 5 search results on Google, and they account for 67% of all search visits. In fact, the top 10 ranked keywords generate over 85% of all traffic from Google.
Once again this demonstrates the importance of first page listings on Google, especially the top half. Page 2 results (rank 11-20) generate only 10% of search traffic, and below page two it amounts to less than 4%.
Again, none of this is particularly novel or insightful, but it’s pretty pleasant to see third party research validated by my own figures. It’s easy to rely on the analyses of others, but you never know if your own sites comply to accepted industry standard metrics unless you go out and actually analyse the numbers yourself.
One thing that did strike me was that each site boasted some anomalous visits from terribly low ranked keywords.
One site received a visit from a keyword where the site ranked 574th, and another site from a keyword ranked 414th. Goes to show that despite the popularity of first page results, some searchers will go to extraordinary lengths to find what they’re looking for.
When next I wanted to analyse the conversion rates of these keywords based on their rankings, I discovered I hadn’t actually enabled ecommerce tracking on these new filtered profiles.
Apparently when you create a new profile for an existing ecommerce website in Google Analytics, the appropriate setting is disabled by default:
So I suppose that particular analysis will have to wait a wee while.
You don’t claim to be a master data analyst so I’ll cut you some slack. While this is a good first step in deconstructing what is going on with a Website’s search referral traffic, you need to take it further and look at:
1) Number of 1st page rankings versus 2nd page, etc. that drive traffic
2) Length of keyword per ranking (that drives traffic)
3) Length of keyword and visitors
4) Estimated traffic volume versus keyword driving traffic
5) Estimate traffic volume per keyword versus 1st page ranking, 2nd page, etc.
Any metric taken out of context is just a random number. The funny thing about statistics is that they can be placed against all sorts of contexts; but denying them any context at all just makes them look warm and fuzzy.
So your charts are only telling part of a story. You want to tell a much fuller, more robust story because inevitably people will look at this and say, “Hey! I DO need to focus on the 1st page for [INSERT VANITY QUERY]” without understanding that they have a lot of 1st page rankings for non-vanity queries.
Hi Michael,
You are, of course, entirely correct. Thanks for giving me material for future blog posts. 🙂
Cheers,
Barry