A Scout is Helpful; this is one of the 12 points of the Scout Law that I grew up learning, and I try to live by it (and the other 11 too). It sure is nice in business if your data partner (aka vendor, supplier) also is Helpful in your relationship and in your data quality improvement initiatives.
It's common (and honestly, it's expected) in the spatial data industry to discover data errors. Over the course of my career, I've encountered a variety of reactions when discussing data errors with various data partners. From these experiences, I will generalize and anonymize the characteristics that distinguish a Helpful Data Partner from a Not So Helpful Data Partner (which is putting it kindly…).
Here are 4 ways to determine if you are a Helpful Data Partner or a Not So Helpful Data Partner:
Do you respond to data issues? How? Do you seem to care, or are issues reported a burden? Data quality issues, whether one-off or systemic, are common. The Helpful data partner uses the data issues as an opportunity for continual improvement and makes improvements to their production & QC systems to prevent future occurrences. One Helpful data partner confirmed our finding, described what went wrong in their production, said they would improve their QC system and then asked for any additional examples of the problems so they could fix them as well. A Not So Helpful partner accepts feedback begrudgingly – if at all. They might blame it on government data sources and abdicate responsibility. They might be so overburdened they don't respond at all. Some of the best ways to improve your data quality and production systems is through insights from customers who have noticed something in your data that you did not.
Do you provide an easy way to report data issues, such as through a GUI that has a tracking system? The Helpful data partner gives you an easy reporting and tracking mechanism for your data issues, including the ability to include your own problem ID in the report so it can easily be tracked. The Not So Helpful data partner has no system aside from email, and might even use a generalized email address to send problems to. Tracking and follow-up is difficult in that case.
Do you provide statistics with your release? The Helpful data partner provides easy to consume statistics on the contents of their product, including deltas (changes between the current and previous release). They also explain proactively if there are large deltas. The Not So Helpful data partner either has no statistics whatsoever with their product or provides only very high-level counts that require their customer to create their own custom tracking systems. Statistics are especially helpful for inventory checks to ensure all the data was properly downloaded and loaded to our production system. They also give confidence in the supplied data when the statistics show reasonable count changes between releases.
Do you report the structural / systemic errors you are aware of, and provide advice on a workaround and an ETA for resolution? Problems happen, and I accept that...but hiding them is Not So Helpful. When systemic data problems are known, the Helpful Data Partner lists them in release notes and includes what release they were known as-of, and when they are expected to be fully solved. When issues are known, they also include possible workarounds, and options for a patched product or query logic that could be implemented by their customers. The Not So Helpful partner either doesn't have any such quality tracking in their production system or doesn't share it with their customers. That puts the burden on the customer and makes us wonder whether we are the QC team of the Not So Helpful Data partner…
Ultimately, this comes down to the attitude of the data supplier. As I write this, I am in a DMV office – a government office which notoriously has poor customer service in the USA – to renew my driver's license. On their walls they have a poster that says, "It's your attitude not your aptitude that determines your altitude." For Data Partners its very similar. Your attitude on data issues greatly contributes to how well your company will be viewed as Helpful, and potentially if your company thrives in the market.What are your experiences with Helpful (and Not So Helpful) Data Partners?
Well said! Quality is a very critical aspect of the data analytics business. with erroneous data, analytics results can be disastrous and cost millions to our client.
Some form of Bad data can come from every data partner. Correct Categorization and scoping of bad data can help in efficient reporting of issues. Let's take a look at a few common categories :
Architect - Data Products, Pitney Bowes Software