I’m an admitted analytics junkie, and I’m going to rant about something I’m starting to see too often from the industry. Why are we now seemingly okay with using filters to cover up inconsistent tracking? Where did the education of our client go, or diligence in maintaining tracking standards?
Within Google Analytics is the administrative function of filters. Filters are these very powerful rule sets available to ‘clean’ the data. When I began using GA, the filters were something to use sparingly and usually with dire warnings accompanying them.
“You’ll never get that data back! It’s gone forever. Poof!”
They were meant to help address the traffic we knew existed but did not want recorded – for example, employees usage, QA traffic when a site has gone live, and spammers. There was a certain brute elegance to analytics filters. Over time, as users increased their familiarity with filters, ie, the types, their functions, their hacks, their usage grew. It was great in that we could create distinct views of the data, isolate audiences and analyze without so much noise and clutter.
Yet that comfort and resulting flexibility has become a panacea and we’ve become sloppy.
The Origin of the Diatribe
My internal fire about filters actually began a few months back. We all have online resources we draw from – you know, those sites you visit when you want to see how someone else approached a problem, or get their take on a topic. You appreciate their insights and opinions.
I was on the site of someone like this who had a post about filters. In there were fantastic descriptions of the filters, what they do, their importance, and so on. And then I saw it – bad filter advice.
The filter in question: the Lowercase filter.
For those unfamiliar, or are wondering what the big deal is with such a benign name, the lowercase filter is meant to take the field identified and force it to lowercase. Available fields include Request URI, search term, ecommerce item name, and campaign (utm) fields. For the purposes of my particular rant, I am focused on the campaign parameters.
Below the fantastic filter descriptions was a call out to a couple of filters that, while available, were not recommended for use. Know what was missing from that list? Yup – the Lowercase Filter. For all of the time we as analysts have collectively spent educating our clients on good tracking practices and data diligence, it just felt as though this filter is a crack in the foundation of analytics governance. “Don’t worry about it, we’ll implement a filter to fix it later…”
Holding Ourselves Accountable
As an industry, we try to educate our clients on governance, data standardization, and data ownership. But I feel we’ve become complacent. I get it, I really do – we’re all pressed for time. Some clients can seem immune to any type of education we give them. But when did we just give up and use filters as stand-ins for tasks such as naming conventions? And become okay with it?
We want to control every aspect of data collection but we simply can’t. Add in that humans make mistakes and you start building a decent case for using the Lowercase filter.
A case for having it to help – sure, go for it. A standard implementation with no other word about education or trying to help clients understand how it’s not a replacement for attention to detail? No, not okay.
Managing Our Clients
Below are some of the frequent reasons I hear from clients not wanting to bother with filter governance (mind you there are others, but these seem to be the most common). This is not meant to counter every point, rather, to give people something to think about.
- Resource hours could be better spent elsewhere.
What about the young marketer just learning about parameters and doesn’t even realize it makes a difference. Or, the young analyst that hasn’t understood the impact to the data of split sources because one was capitalized and another not?
- Too many people/vendors/companies involved to try and get everyone to change.
Standards aren’t meant to change all of the time and can take some time to roll out. Consider implementing any necessary updates as new campaigns roll out. Or better yet, this would be a great exercise for someone to spend time learning a new system or a fundamental of tracking and analytics.
- But then it won’t match the old data.
Sure it would. That time you had your analyst manually calculating stuff because of the sloppiness? That’s going to continue until your time periods of analysis become clean. But then it will be faster.
- It’s just part of the standard set-up.
It is, now. We are an industry which espouses the importance of details, even when we look at data in aggregate. We want the number of transactions in a month, but we want to see it broken down by GA UID. Why would we not do it right to begin with, and educate our clients to be aware of the details and impact of sloppiness?
- It’s just not a big deal; the filter will fix it.
You’re missing my point.
I’m not saying that the Lowercase filter doesn’t have its uses. Not every implementation is going to be as clean as we hope every single time. But we still owe it to ourselves, and our clients, to maintain a bit of data integrity. Be mindful of what you recommend to clients, and help them understand the longer-term ramifications of their decisions.
Acknowledgment and Buy-In
Are willing parties going to partake? (Really, is everyone involved going to adhere to the standards? All too often one department creates some great standards, and doesn’t understand why other departments aren’t just as excited when – surprise! – the standards are foisted upon them.)
Defining and Introducing Standards
What is the most compatible for the most vendors in your network?
Any specific system considerations? (one example being paid search in which capitals may be used in the ad group name, which later is forced to lowercase)
Does this change any perceptions of the data? (I had a client who was using capitalization to determine who implemented each campaign – e.g. source of Facebook versus facebook.)
Are URLs going to be checked prior to campaign launch?
Is there going to be a defined QA process?
Will someone periodically check reporting for tracking cleanliness?
An older post, but one which I think the key steps are still applicable to consider: http://marketingland.com/2015-plan-needs-include-data-analytics-governance-105962
And while I won’t say which of the sites started my diatribe, there are a few I’d like to call out for their great information over the years. I still trust them and they will continue to be my go-to resources, and I suggest you check them out, and even bookmark them, if you haven’t already:
But what are your thoughts? Much ado about nothing? What are some blogs that you love, or broke your heart?