I’m using Google Analytics as a metric to provide quantitative data towards an Action Research cycle.
I get pretty good web logs and working awstats from my web hosting provider for distributedresearch.net, but back in August I decided to try adding the Google Analytics package as part of an Action Research approach towards improving the effectiveness of my blog.
Action Research is more than just testing and tracking, but the testing and tracking methods used by professional internet writers, bloggers, developers and marketers begin to look quite similar to distributed action research in some aspects since they are being applied in complex adaptive situations where pure quantitative metrics and ideas such as a ‘fair test’, split testing and testing against a control set are mostly inapplicable.
The specific and limited aspect of the blog which I have been trying to improve in this way has been the attraction of new visitors to specific posts via the search engines. So I added the Google Analytics code, that’s one action in itself and a very powerful one, and I adopted a more consciously search engine friendly writing style for a series of posts. The research programme ran for about six weeks but during that time I was also making other adjustments to the overall performance of the blog. So there is no way that I can attribute any increase in traffic as being entirely due to one specific action, but that is OK because my goal is to improve the reach of the blog, it isn’t to prove a point. Besides, when you are dealing with unknowns such as the Google ranking algorithm or the interactions between thousands of publishers and tens of thousands of casual web surfers then nothing is either constant or predictable. A page can drop from position number 5 on the first page of search results down to number 13 over the page for two days, and then return at position number 3. The amount of traffic thus fluctuates wildly for those few days with no action having been taken on my part. The totality of all the variables involved is unknowable in a complex adaptive system, but that doesn’t mean research is impossible. You can still prod and probe and gain actionable insights, that’s what action research is all about
Where Google Analytics helps is in the automatic generation of trend tracking graphs for multiple slices of data, which is just great for research purposes. I can look at the traffic which arrived via search engines for the period involved and see a nice upwardly moving path.
So some of my actions may be working well, and I can drill down further to visualise traffic for specific posts, see the most popular pages, aggregate data for individual keyphrases and then zoom back out to see the overall picture.
Those little saw teeth are spikes caused by StumbleUpon by the way. The temporary traffic increase from that route is fleeting and not sustained, but may pick up one or two more interested readers, you never know.
The optimised posts combined are steadily gaining attention according to the stats, but what are the limitation of this approach?
Limitations of Google Analytics
Google Analytics provides only quantitative data. For Action research to provide a full picture you need to collect qualitative data, which would means gathering narrative from the visitors and readers – not an easy task with distributed action research based around search engine traffic, but not impossible either. Any ideas posted in the comments to blog posts for example would count as valuable qualitative data. Watching over a friends shoulder as they search and surf is another way to gain insights as to some of the alternative habit which different people adopt. Then there is all the data in my RSS reader which comes in as a mixture of advice from subject experts and experiential narratives from fellow learners.
Action Research requires a process of reflection between cycles. This prevents the process from running away on its own momentum, provides a check that events are moving in a direction which is in line with stated goals and values, and offers an opportunity to asl wider question which may reveal deeper insights leading to a positive change of direction for the action research project overall.
The graph of traffic via referral causes me to pause. In pursuing one goal, search engine traffic, am I leaving some previously regular readers behind. That’s a known risk, but I wonder to what extent there may be conflict between subtly optimising for search engine traffic and providing value for regular readers. I hope to have kept any such conflict to an absolute minimum, but what do you think? Can I realistically spend two months working on one type of traffic, then just switch over to a different focus for another period, perhaps engaging with regular readers more or attracting RSS subscribers. Or could it possibly be a better approach to try and combine everything at once, always considering every aspect?
The four posts which have been attracting most search traffic are