The Secret to an Effective Data Quality Feedback Loop

As ever, Tom Redman ("The Data Doc”) recently prescribed some great advice in his latest Harvard Business Review post: "Break the Bad Data Habit”.

It’s great that Tom is reaching a new audience with his publications on HBR and one of his recommendations is often overlooked by most organisations I visit - the need for data quality feedback.

How many times have you spotted an error in a spreadsheet, report or email and applied a local fix? Did you pass details back to the source department? Was it even possible to find the source department?

When we’re busy going about our daily working lives we’re handling data all day long so if we can avoid unnecessary administrative chores we’ll often take the path of least resistance to our workload.

A personal example of poor data quality feedback

We recently bought a new family car and the salesperson entered an incorrect date for my wife on the registration form. He quickly scrubbed it out with the correct one but then additional paperwork arrived, all with the incorrect date.

Clearly he had not bothered to contact central office and get them to update the original records so the error persists.

Lack of a data quality feedback loop invariably leads to downstream issues. In the car purchase example we now have to contact the dealer again, which is frustrating as a consumer and adds more time to the central office administration team. If the error had been fixed in all the source paperwork then all of this could have been avoided.

In Tom’s post he also talks about the issues of accountability and this goes hand-in-hand with feedback. People need to take responsibility for their own data so cultural maturity is important here also but this can take time and it can be hampered if the feedback loop is complex and unwieldy. If someone has to complete forms and change requests to correct a defect they’ve spotted then it’s unlikely to happen.

Data Quality feedback loops needs incentivisation to work

I used to see this problem a great deal in the Utilities and Telecoms sectors. One site survey I carried out found that nearly 40% of equipment on one floor in one exchange was incorrectly recorded. One of the contributing factors was that engineers were not sufficiently incentivised to correct data. Their measures of performance were based around speed of completion. The irony of course is that defective data was the primary cause of increased lead times on provisioning and service activities!

Applying the "Cookie principle” to data quality improvement

A great example of this was in our earlier interview: "Going beyond Six Sigma: How KFR Services Inc. took their data to a whole new level of data quality” in which customers were incentivised to use the data quality feedback loop and report defects in return for cookies!

You may also notice that the data quality practitioner who helped get their program moving was none other than Tom Redman, proving once again that simple ideas, executed correctly, are the key to data quality success.

So to create an effective data quality feedback loop we need a process that is:

  1. Simple in execution, to guide people step by step
  2. Incentivised, so that people feel rewarded
  3. Time sensitive, having a simple process is important but it needs to also be fast

What does your data quality feedback loop look like?

If one of your data customers (either internal or external) spots a data defect what steps does your feedback loop consist of? Is the process easy to follow and well maintained? Do you get positive comments and regular activity? Where could it be improved?

Welcome your views on data quality feedback loops and how they’re helping to improve your data quality initiatives.

Photo credit: http://farm1.staticflickr.com/112/271751746_01d1ac8b33_m.jpg