Ecommerce data is growing exponentially. The latest push is driven by the effects of the pandemic, but we have been on this path for a long time. As more and more transactions take place online, companies are gaining an unprecedented level of customer insight through the trail of data left behind. However, this is leading to major pitfalls for companies without the right analytics teams, data processes, and data governance in place.
When the RXA team starts an analytics implementation, we frequently see massive, missed opportunities with data collection. The one place that data integrity should be pristine is often the most cluttered and problematic: the eCommerce website. The data quality coming in from websites varies wildly. We see GA tags (Google Analytics tags) and Adobe Analytics tags missing or duplicated regularly. A simple tag management system, like Google Tag Manager, does not offer enough information to catch all mistakes and ensure accurate data.
Tag Manager Pitfalls
There was a recent incident where website activity for a customer appeared to grow by over 800%, and this coincided with a new marketing launch. The team was initially excited; however, purchases and leads did not increase. Turns out, this was not due to thousands of new customers coming to the site, but their agency had accidently added duplicate tracking events to the landing pages alongside the new marketing campaign tags.
It took 3 weeks to clean out the bad tags, and we now must account for this window of bad data forever. We were able to adjust this period of data in their Domo implementation, so their marketing analytics team doesn’t have to constantly work to remove this inflation. But these are issues that often go unnoticed. We were actually lucky the issue was so large; it was obvious to anyone (other than the people who wanted to believe in the miracle marketing campaign) that this was a tagging issue. But what about when it’s not obvious? Or when you are using bad data to forecast customer churn, spending marketing dollars to fix problems you don’t have, or not spending to fix problems you do?
ObservePoint and Automated Testing
When the RXA consulting team is setting out to implement data science applications, we can spend up to 80% of our time working on cleaning data, performing custom ETL (extract transform and load) tasks, and other activities to ensure we have good data. But it’s always better to get good data upfront, and not have to go back and fix it. If you are building a forecast and the input data is bad, and you don’t know it’s bad, you now have a compounding problem that you don’t even see.
We have had success using tools like ObservePoint that monitor website tagging and highlight errors. ObservePoint offers a product called Journeys, which automates testing and monitoring of user journeys on your website. You can specify a critical user path and ObservePoint Journeys will automatically follow the flow and make sure that data is collected accurately along the path. It’s that simple. No more manual monitoring, and no more missing or inflated data.
As long as eCommerce data exists, users will make mistakes in data collection and analysis. We are not perfect, but the great thing is, we don’t need to be! Tools like these allow us to catch mistakes, large or small, before they alter business decisions.