When it comes to my work, I have always been a very process oriented person. I find that this serves me very well when it comes to Web Analytics projects, especially in the implementation and testing phases. I’m not alone in finding processes helpful to keep all the complex steps in implementing web analytics getting completed. I know of at least one enterprise analytics company that has developed a very thorough process, and it’s not dissimilar to the process I personally use with my clients.
Let me walk you through a typical implementation from a high-level and you will clearly see why having a detailed process in place before the project begins, will ensure everything is completed on-time.
With any web project, the first step must be gathering requirements from the client. Before I can determine the best KPIs (Key Performance Indicators) to track, I need to first understand what the clients’ goals are for the site. Many of the sites I have worked on have been micro-sites with a very clear, specific goal (contact an agent, download a PDF, sign-up for a newsletter). However on larger, more complex sites, there may be multiple goals, and therefore more complex KPIs. I do try and keep the KPIs to 5 or less, no matter how large or complex the site is. If this isn’t possible, I try and consider breaking the site up into sections, and identifying 1-2 KPIs per section.
Since I work with agencies, they often do this part of the process without me and just send me the documented requirements.
Once the requirements have been gathered and the KPIs have been determined, the next phase I always do is put together a detailed tracking manifest. This is a spreadsheet with a list of every page/screen of a site, and every track-able action. I often go through and list every single action and then trim out the detail that isn’t needed. I feel it’s better to list too many actions, than to accidentally miss a few critical ones. Within the manifest I detail out things like the page names, type of action (onLoad, onClick, onSubmit) to be tracked, and any and all custom variables it makes sense to set for each.
Depending on the client, either I go ahead and do the implementation myself, or I work closely with a developer who will be doing the actual code implementation. In either case the detailed tracking manifest becomes very critical at this stage. Many sites are complex and it can be easy to miss adding the code for a specific custom variable if there isn’t a detailed check-list to help guide you page-per-page. This ensures all the necessary code and code snippets are not missed.
Quality-Assurance (QA) Testing
The final part of the process is doing a thorough round of QA testing. I use my tracking manifest as a test plan as well, and go through every page and action to make sure each step is being tracked originally intended. This requires the site to live in a staging server that is tied into the analytics tool. I like to use a combination of tools to test the implementation. I have a variety of FireFox plugins, a couple of stand-alone code debuggers (Charles and Fiddler 2) and the analytics tool itself that I look at during a QA test.
So as you can tell any analytics implementation can be quite complex, and without a detailed process in place, things can quickly go wrong and delay or even derail the project. Make sure you have your ducks in a row ahead of time and you will increase your chance of a smooth and successful implementation.