The question, ‘Is your analytic data performing well enough?’ is one that lays a lot of emphasis on the fact that analytics is an entire field of study. It is not a process that ends with a few steps, because output of one stage of analysis serves as an input for the other stage of running a business.
Moreover, there is a high chance that you may select the right metrics, and yet you fail to analyze them properly, or that you may have all the analysis in perfect order, yet you fail to implement the findings just right. Hence, there is a lot to be monitored in web analytics because an error on any stage can lead to underperformance of analytic data.
Before getting into the ‘how’, let us make the ‘what’ very clear. Deciding whether analytic data is performing well or not is a question that is preceded by the correct identification of the purpose behind the analysis. What do you want web analytics to achieve?
Broadly speaking, web analytics is all about tracing the cognitive trail left behind by visitors. Dissecting and exploring the intents and behaviors of the visitors gives insights regarding their actions, and more importantly, why they behave as they do. Hence, if your data is only figuring out what the visitors did, instead of telling you why they did it, it is far from yielding results.
In a nutshell, the purpose of the analysis should be clear to you, since it is based on this purpose, that the efficacy of the gathered data can be judged.
What will the analytic data search for? Treating the entire web tracking process like an extensive research program is the right approach at this stage because, before the data is gauged for effectiveness, it has to lay down the research questions.
Once you have identified the purpose, develop a research question that is more detailed, identifies the aspects of a website that will be looked into, and explains why this research is important, i.e. what will it yield. Everything about web analytics will then be geared towards answering this question.
Many webmasters prefer to jump straight to analysis instead of laying down these basics. Even though such an approach may work for them, it takes away the formalization in the web analytics process that gives it the importance it deserves.
KPIs and metrics
The third criterion for data to perform well is to have in place the right KPIs and metrics. Both of these topics have been discussed in a lot of detail in the previous chapters. They are essential aspects of web analytics because they give practicality to the research question. Once these are identified and documented, data can be gathered with certainty because all performance indices are specific to the type of business and website under the spotlight.
Every metrics and performance indicator should be able to track four types of data:
1. Source – Source indicates where the prospects come from. Do they find your website from search engines or are you referred by another website?
2. Amount – Amount determines the number of instances of a particular action performed by visitors. In other words, it takes into account the frequency at which the action is performed.
3. Nature – What is the nature of the activity? What do the visitors experience when they land on your website or webpage?
4. Results – What action is taken on the website? Metrics that determine results or goals are highly important because they indicate how well the efforts of a business are reaping rewards.
The last criterion to determine the performance of the analytic data is benchmarking. Even though this has been talked about previously, it is important to mention benchmarks in this chapter because they play a huge role in determining success.
When a webmaster sets a benchmark, all data gathered suddenly starts making sense. On their own, most metrics are absolute numbers or percentages that do not give any results nor do they show how the data gathered is of any good. Once a benchmark is set for a metric, the level of good or bad becomes clearer and the analyst can then determine how the data is performing.
With benchmarks in sight, experiments and tests can be conducted to reveal how the website is performing. Leaks in the conversion funnel can also be identified better when metrics are gauged against benchmarks and the resultant gap is then analyzed.
Apart from these four measurable criteria, a lot of the responsibility for the performance of the data lies on how well each aspect of the analytics process is knit together. Integration in the process and design is crucial for web analytics in particular, because the results derived are a summation of all metrics and tools used.
It also helps to remember that performance is a very subjective measure. There is no ‘right’ level of performance because it varies depending on the industry and sector. Apart from the industry, a lot of other factors affect the efficiency of analytic data, including the tools used, the approach taken, the difficulty of assessment and the time horizon in mind.
Therefore, while you may want your data to achieve an optimal conversion rate, the definition of ‘optimal’ is not as absolute as one might think.
There are a number of professional and specialized web analytic companies that have trained experts who have web analytics and all its jargons on their fingertips.
However, not every business has the financial backing to pay professional web analysts to take a look at their websites. Nor do they have the necessary infrastructure needed to accommodate an external analytics website into the reporting and monitoring framework. How can webmasters of these companies benefit from web analytics?
The beauty of this process is that it can be implemented with a little knowledge about metrics, data punching and benchmarking. This is not to say that web analytics with professionals is no different. When a specialized company is contacted, website assessment takes a completely new meaning altogether. However, until your company builds the muscle needed to get professional services, using testing software is the best way to make use of web analytics at minimum costs.
Two testing practices in particular make it to the top in regards to web analytics. These are:
- A/B testing
- Multivariate testing
Both of these testing methods are used for conversion rate optimization because they lay the ground for a webmaster to experiment with different options, assessing the effectiveness and efficiency of each. Depending on your needs and the level of analysis you want, you can take your pick from either of the two as the first step towards analyzing your website and evaluating its performance.
In the context of web analytics, A/B testing requires webmasters to create two versions of the same website or a webpage that is to be tested. Both versions are differentiated by changing their designs, layouts and other major characteristics of a website. Next, both websites are given different URLs and traffic is divided between them to test the effect of the changes.
Using this testing software, any number versions of a webpage can be created, depending on how much research you want to do regarding website performance. A/B tests can be changed into A/B/C if three versions are being tested and even, A/B/C/D if four versions need to be considered simultaneously.
An A/B test has become very common nowadays and it can easily be implemented by most webmasters who want to go ahead with analytics on their own. A/B testing can be carried out using a number of software, many of which are available online without any costs.
- Conversion rates are not restricted by a particular area of a webpage.
- Drastic changes to website design, layout and consolidation can be made.
- Advanced analytics skills can be incorporated with the simple version of A/B.
Multivariate testing goes a step further to analyze the effect of different variables within a single webpage. It helps a webmaster change the placement of call to action buttons, headers, forward links and many other elements of a web page to optimize its performance.
The multivariate testing is more advanced than A/B testing because it reveals how the performance of a website relies on specific elements that are already a part of it. The combination of these elements and their optimal usage is another result a webmaster extracts from this testing technique.
Just as A/B testing can be done using various software solutions, multivariate tests can be run on a number of platforms. Visual Website Optimizer, for instance, offers a detailed analysis for this test, together with idea-generation tools that provide the best starting point to run this test.
- Isolation of many website elements is possible. Understanding their individual effect becomes easy.
- Individual effects can then be compiled in an organized way to study compounded effects of all elements in unison.
- Conversion rate optimization is stricter and the criterion used to gauge it is very narrow.
Advanced web testing procedures often prefer to use both these software to arrive at very detailed and lengthy results. Using both tests enables analysts to view a website’s performance under the microscope as well as from the standpoint of a general internet user who merely visits once or twice. Keeping both these perspectives in mind presents a better solution by testing page layouts, as well as change in variables, all in one go.
Choosing either one of these testing procedures becomes easy if you have a few essentials lined out. For instance:
- Very high traffic on the website
If your website receives high volume of traffic, using multivariate testing is sensible because the minor changes in the variables of a website will then be noticed by a larger audience and measurable results will be gathered.
A/B testing, on the other hand, can work on web pages that receive little traffic because the analysis is more superficial than that generated by multivariate testing.
- Quick results
Do you want to do a quick test or have an in-depth analysis of the website under consideration? A/B testing usually gives quick results because it can be conducted without making major changes to multiple variables. If you want a relatively simple overlook of the way your website would work best, consider A/B testing first.
- Interaction of webpage elements
Since multivariate testing goes further into the working of various elements of a webpage, it tells a webmaster how these elements interact with each other to optimize the performance of a website. If your goal is to find data on variables, this testing software should be your target.
- Average technical know-how
If you have an IT team that has average web analytic skills, it is better to start with A/B testing because this procedure requires a relatively simpler framework as compared to multivariate testing.