I am not a huge fan of Google Analytics…I’ll say that to start off with. I might write about my reasons some day, but for now, my bias should be known.
Yet, it is a free tool, an almost universally used tool, and thus it’s worth talking about to help make sure people get the most out of it.
Of late, I’ve been training a new Analyst on our team at Delphic Digital, and it’s reminded me of something I haven’t thought about in a while. Namely, how important it is to understand and relate how you implement GA to track your site’s behavior to how it will be reported. GA doesn’t have a lot of flexibility when it comes to report configuration or metric calculation. Maybe it would be more truthful to say it has flexibility, but a lot of times that has to be done through multiple Advanced Segments, Custom Reports, and so on. Not the fastest thing to do when you are digging around to figure out where the key to the question you have is.
Take for example Goal Conversion Rate when you’re looking at it by traffic source / medium. This sums the Goal Conversion Rate across all goals. On first blush, this makes sense. However, let’s say you’re an ecommerce company with a checkout flow and set up Goals for each or most steps of the flow to understand flow, but also to give value to each of those steps. (That’s not something I necessarily recommend doing, but companies do it all the time.) So, then when you’re looking at Goal Conversion Rate by source / medium, and you have two dimensions that have relatively similar values for the Goal Conversion Rate metric, you need to figure out whether they really are that close, or is there more of a story. The trouble with Goal Conversion Rate as a sum of all the individual Goal Conversion Rates is that it doesn’t let you differentiate the value between a 5% and 6% overall Goal Conversion Rate if some of your Goals are more valuable than others. So you might say well then just look at the Goal Conversion Rates for the goals that are more valuable to you, or at the Per Visit Goal Value for that matter. (If you are properly setting your Goal Values, then the Per Visit Goal Value would give you the differentiation in value of different Goals.)
The trouble with the former is that if you have 20 Goals (and which substantive site doesn’t want to have more?), then you have to creates sets within that 20 each time you are doing such analysis. And I generally find that the 4 goals per Goal Set thing doesn’t end up working out well when you are running analysis for different purposes. (i.e. sometimes grouping Goals one way will make sense for UI analytics, but grouping Goals another way will be better for Revenue or Product Merchandising analytics)
I do love Per Visit Goal Value, but I do find that it often is not that accurately set and it takes time to get the Goal Values relative to each other right.
So, this all leads to me say that when you are setting up your Goals, think about how they will be reported and viewed. For example, if you don’t need 10 goals, don’t use them, because more Goals means that the overall Goal Conversion Rate will be tougher to break down. It’s not that breaking it down is impossible, but that it takes time. And time is something we often don’t have when we get “an hour of time to run the numbers on such and such change or campaign”. And specifically relating to Goals, I recommend only implementing the most important or telling “conversion points” on your site, and let your data flow for a week or two before seeing where you need to add more Goals. One of the reasons is that if you ever want to get rid of a Goal, you don’t have a way to flush that Goal of it’s data. So, if Goal 1 was reaching Page A, and now you’re not interested in that anymore, you can’t flush the meaning of Goal 1 as Page A. You have to simply annotate that on such and such date, Goal 1 changed to being some other day. I recently had to deal with this when we had a large site completely relaunch, and almost all of the 20 goals were meaningless now. That makes for fun YoY comparisons 🙂