Once established, analytics data quality tends to be taken for granted. The site is tagged, the filters applied and the data is flowing in.
Battling the metrics then begins and the mind wanders off from auditing the data, but nothing can ruin your new staggering insight quicker than that dawning realisation that the data is flawed.
So many elements can affect the data flow. Sites change, pages are renamed, structures altered, processes amended, tags change, offices can change physical locations.
All can cause issues with implementations that may even have not been configured correctly in the first place.
As Ben Davis reported in his recent blog post on the cost of bad data, data quality issues cost businesses on average 12% of its revenue. Read it. As a Data Marketing Manager, it’s terrifying (and a little bit shameful that I recognise a number of these issues).
Google Analytics has introduced its new Diagnostics feature (currently in Beta) which serves to help you identify and fix some of these issues.
It says it’s to ensure you can continue to “fuel powerful actions like improving websites, streamlining mobile apps, and optimizing marketing investment” – which it surely does.
I think it more fundamentally tidies up some rather disjointed tools provided in the past to support the analytics administrator. Hands up if you have ever experienced:
- Lack of trust because a report you sent through six months ago was later shown as being flawed because of some data issue, and subsequently all reports have been tainted.
- Page changes destroying your carefully constructed funnels.
- New colleagues or agencies asking why simple configuration changes had not been applied to an account in front of a boss or client.
Now I’m not suggesting whistleblowing is bad, far from it. I just like the idea that these configuration and auditing issues are quietly identified quickly and concisely. And to me first – all in one place.
As ever, Google is a bit fuzzy in what Diagnostics regularly scans for, suggesting site tagging, account configuration and the data itself are inspected, mentioning specifically:
- Missing or malformed analytics tags.
- Filters that conflict.
- Looking for the presence of (other) entries in reports.
For the diligent analytics administrator, Google’s Tag Assistant for Chrome, and carefully crafted Intelligence Events within Analytics itself would have covered a selection of these issues, hopefully the beta period for Google will allow the scope of these identifiable issues to grow as this really does have the benefit of being a really useful single tool.
The Econsultancy site did not pass unscathed (although not as badly as it could have thanks to the army of people ready to identify tagging issues and other problems on our site) – a number of site changes have left a selection of goal funnels without information.
This is not necessarily an issue but again points to the quality of the information you are allowing the business to see. I’ll be busy over the next few days…