The following is Part V of E-commerce & Web Analytics:
Part I: Visitor Targeting & Acquisition
Part II: Landing Page Effectiveness
Part III: Cross Selling & Checkout Optimization
Part IV: E-commerce Segmenting

Modifying & Testing

Once data has been gathered, it will need to be analyzed. Analytic programs capture, store and arrange data, but they are not capable of the analysis. This is the human element in analytics. Theoretically, the data collected by the analytics program is always correct. The program is not capable of producing erroneous data, though it could fail to capture some data due to bad tags. Assuming it does collect all the data, there is no questioning the validity of the data. What needs questioning is why that particular data was collected and to find ways to produce data that reflects the wants and needs of the E-commerce site.

Through analysis, hypotheses are born. These hypotheses will need to be tested, measured, and reanalyzed. This is the role of the analytics team. First they must analyze the data collected for trends, as well as direct & indirect relationships. This analysis will lead to conclusions that may or may not be true. Thus, a hypothesis is born and now needs to be tested. Once the changes are made, the analytics program captures the new data produced, it is compared to the data pre-hypothesis, and the hypothesis is proven true, proven false, or is inconclusive. Regardless of the result of the testing, it is the testing itself that is most important. There will be many a hypothesis that is proven false, but it is not time wasted. It simply proves that a relationship exists that is different from the one the hypothesis presumed.

Without modifications and testing, websites become stale and stagnant. Even tests that produce adverse effects have value in that they show a relationship that previously was not known. Clearly, had the analyst known that the changes would produce detrimental results, he/she would not have suggested the changes. So in failure, knowledge is gained. Likewise, when modifications produce positive results, knowledge is gained. Whether X produces more of Y or less of Y, we now know something about X we did not know before. This new knowledge can be developed and implemented if good, or it can be noted and avoided if bad.

The role of the analytics program is not just to collect data, but to measure changes in data. The more powerful the program, the more ways the data can be measured, stored and tested. This flexibility allows for deeper analysis and greater revelations. The data created by visitors is all there for analysis, but the power of the analytics tool determines how much can be seen.

Repeat

This section is fairly self explanatory. The process of collecting data, analyzing it, formulating a hypothesis, making modifications, and testing the hypothesis should be in perpetual motion. There is such a wide array of factors that induce conversion and no one hypothesis can account for them all. Rather, it will be a series of hypotheses that lead to small changes. Each small change could have a small or large effect on conversion, but it is the sum total of all changes that will bring prosperity. Sometimes it may be a matter of taking a step back before taking a leap forward, but it is the willingness to take the step in either direction to learn something that is most important.

The best E-commerce sites are constantly being modified and tested. They are not afraid to test and make modifications to portions of the website that are already working well in the hopes of making them even better. Naturally, one should first concentrate on areas that are shown to be working poorly or not at all. However, the only way to prove something cannot be done any better is to test it and show any or all modifications to it prove to be worse. Just because something is working fine, does not mean it cannot be made more efficient.

Summary

The quality of the analytics tool can affect every level of the E-commerce conversion funnel. From acquisition to cross selling to conversion, the analytic tool should facilitate each stage of the process. Analytics allow for the creation of metrics which can then be tracked and measured for changes. The changes in these metrics are the breadcrumbs on the path to profits. The better the analytics program the easier the path is to follow.