More data is usually better. But sometimes you get so much random data that what really matters gets lost in the noise. Badly formatted data can be worse than too little data − you have it but can’t figure out what it means.
Modern digital analytics can track almost any action on a web site or app. Basic tracking is easily done by following simple instructions on a wide variety of tools. Going further, skilled analytics developers − such as the technical analysts at 6D − can implement code to track just about anything.
Let’s talk about how the way in which you implement analytics affects the data quality.
A typical analytics implementation starts with page traffic. You get this from just placing the basic tracking tag. Any analytic tool gives you page/URL tracking. After that you determine which events to track: link clicks, forms, on-page interaction, etc.
This is where it can go wrong. The question of which links to track gets a response of “Track all the Links!” Which buttons? “Track all the Buttons!” How about … never mind, we’ll track all the things. (Heavy sigh)
When we actually try and use the data, we can’t correlate a specific button click with results. Dynamic site optimization programs make it worse when yesterday’s “Buy It” button is now today’s “Purchase” button, but 10% of the visitors have a “Get Me” button.
- Increases the amount of noise data you have to sort through
- Can cause errors not only in the analytics but for dynamic page content
- Bloats the size of analytics code and how long it takes to run
- Increases the effort needed by analytics and site developers
Important points in analytics tracking:
- Track what you need
- Have a tracking strategy and a plan
- Regularly check data and update where needed
Track What You Need
Track what you need; don’t include everything just to be gathering data. To do this you need to determine what you need. With modern tag managers it is easy to add additional tracking later.
Start with site KPIs and site goals. If you don’t have any, get some. We’re not going to cover that here other than put some thought into what you want your site to do and how is it supposed to be accomplishing that.
Do the items you are tagging support those KPIs and goals? Directly, not “maybe”. This helps you start your “to be tracked” list. Directly link, “this event lets us track this KPI”.
Is there a business reason for this data? Is there someone in sales or marketing who will be using this data to make decisions or perform business actions? If not, why are you tracking it?
Is this data going to be used for site improvement − directly and immediately? “I might use it in the future” is not a reason to track something. If we need something in the future we can put in the tracking, then remove it when not needed. Related to this, use some type of tag manager or governance system that enables you to add and remove tracking features as needed.
Basically, if you don’t have a user for the data, consider not tracking it. Most analytics implementations list the “What” (data) but neglect the “Whom”. Think of the analytics user.
Tracking Strategy and Plan
Too often, an organization does the KPI portion (all business training tells them to) then just starts throwing tracking code at the site. This gets bogged down in creating separate code for every button/link/action on the site. This increases the implementation effort and makes your data as scattered as the implementation.
Organize the tracking into categories. This starts to diverge from the KPIs to become the action type, and it should relate to the type of data and user. For example:
This may be a bit controversial. Your site or analytic code developer will tell you things should be organized into event types: page, link click, object interaction, etc. Don’t. A key point is that tracking code should support the data use, not the tracking code developer.
If your site is more focused on business groups use those as a top level with the function underneath.
- Red Widgets Group
- Funny Shaped Plastic Things Division
- Unknown Objects Development
- Corporate Overlords
Sort the elements into their use type. Having a data organization plan is more important than the specific organization.
Then have your site and analytics developers sort these into common elements. All “Buy Now”-type links should have a common way of tracking. They should have a common naming convention so that the data groups are identifiable and roll up into their end use, and won’t break if there is a minor link text change.
I won’t cover the coding details. By this point you should have the group, action, type, etc. details worked out so that this becomes the requirements for the specific event tracking. This saves time as you do it across the site level instead of doing this evaluation for each separate event.
Audit and Update
Your site never changes so your analytics data configuration will never need to be changed, right? And I’m still using my GeoCities site through an AOL dialup. Hopefully your site does change as you improve it and add new content. A static analytics system will break over time as the site diverges from the one on which tracking was originally implemented.
Your analytics will need to be part of site changes. But we created a system where the final code falls into an organized structure! Aren’t we lucky we did this and documented it (you did document it, didn’t you?)
Create a process where site changes get analytics added per the tagging strategy. This should be minimal effort because most of the definition work is already done. Just fit the new events into that structure. Modify the tagging strategy as needed, so your site and analytics are a growing, adaptive structure.
Audit the data regularly to make sure that any step changes are integrated (e.g., paid search-attributed conversions drop 90% but you haven’t reduced search spend). Regular data audits will help you see that something has changed that may need an analytics update. This will happen − if not from you missing changed content on your site, then from one of your traffic sources changing a parameter you were tracking.
Finally, circle back with your business users regularly to make sure that what you are tracking is what they are currently interested in. Use a tag manager to remove tracking not being used and add new elements.
A quick summary of successful and maintainable analytics:
- Avoid tracking all the things − concentrate on the important things
- Organize the tracking so the data is usable and maintainable
- Regularly check your analytics against reality and update as needed
Curtis Smith is a Senior Technical Analyst with 6D Global.