How often have you not acted on what your digital analytics are telling you? How often is that because you don’t trust the numbers? Data accuracy, or something like it, is key to data relevancy.
And I do mean “something like it.”
In an environment where you are counting millions of data points, total accuracy should be a non-issue. Wasting time on the achievement of total accuracy is as much of a time-suck as trying to make sense of very inaccurate data.
What is “accurate”?
In the real world, it’s an artifact of rigor. You cannot know the data is accurate. You can only know how much effort you’ve put into making it accurate. And if you’ve put serious effort into it, then you are as close as you’re going to get to true accuracy. An old maxim in the data business is that the last 15 percent of accuracy costs as much to get as the first 85 percent. It’s up to you how you want to spend your time and money.
So, let’s ask again: How often have you doubted the useful accuracy of your data?
Put another way, how do you get comfortable enough with your data that you can risk trusting it?
There are a number of ways to make sure your data is accurate.
The analytics engines do their work behind a wall and you have no control and haven’t any idea whether they are just making stuff up or not. But because that matter begs too many questions that go well beyond the purview of data itself, we’re not going to do more than mention the possibility.
Your job as a marketer is to carefully review data collection.
You can control data collection. You can also control the range, look, and feel of reporting, but none of that has any impact on accuracy. To get comfortable with your data, make sure you spend time checking data collection.
Data collection is largely a matter of what now is commonly called “tagging.” The term has become so common in digital analytics that at times it seems as if everyone knows what it is and how it works and so if you are a tag-invested marketer already, skip the next paragraph.
Tagging is accomplished by several lines of mostly standardized code, provided by the digital analytics vendor and often customized locally, that gets placed into the html of a page usually but not always in the header. This code communicates with the vendor’s data warehouse, carrying a who-what-where-when story about that page and its visitor. That data gets processed by the analysis engine and the analysis engine delivers formatted data to the visual reporting layer that you’ve become comfortable with over time.
It sounds fairly straightforward. But still you doubt the accuracy of your data.
Ask yourself if you’ve been rigorous enough about data collection. Who placed the code on the page? Did they work with data collection experts to make sure it was placed properly? If they did that, did the experts provide you with a tagging specification that could be deployed across your digital properties? Even though it sounds dumb, did you check to see if every page actually got tagged? Did your experts dig into details like cross-domain tracking? Proper placement on the page? Variables properly set? Too many “load items” on the page? Too many instances of the code? Latest version of the code (which can change fairly often)?
There are more things to check, but we haven’t the space here to go that deep.
Who Are You?
Let’s say you fall into one of three categories:
- I did all that and am comfortable with my data
- I am not comfortable with my data and need to do all that
- I was comfortable with my data until just now
Here is the call to action: unless you are in the first category above, do all that. Then, you can lie awake pondering another question: What is the value of a good night’s sleep?
Image via Shutterstock.