Through the many interviews I’ve carried out with experts on Data Quality Pro, I’ve learned that the biggest data quality challenge is simply getting started.
We all know that data is important. Even the most stubborn sponsors agree that information is critical to business success, yet time after time projects are side-stepped or put on hold, often indefinitely “until the time is right” or the “financial situation improves” or “we get the right resources.”
Why hasn’t the start-up problem been solved?
A common problem is that the vision is just too grand for a sponsor to visualise. Data quality is typically a greenfield endeavour for most sponsors, so why would they sign up for something so risky without proof it actually delivers?
If there is doubt or confusion, it breeds risk and fear - never a good recipe for sponsor momentum. As I pointed out in an earlier article on the Roundtable, one of the principle issues is changing habits.
How can you approach this problem differently?
I’m a big fan of the guerilla approach to data quality and governance. That’s not to say I think everything should be bottom-up - I still think the top-down leadership model is critical. It’s just that I think that when faced with open blockading of your vision and quest, it pays to think in a more organic, agile way.
Do you really need to pitch the entire vision and strategy for data quality right from the outset? What if there was another way?
Conventional wisdom often states that you should build frameworks, methodologies, expert teams and all the associated plumbing and financial infrastructure that goes with this capability.
I’m not convinced. In fact, in my most successful initiatives, we started very small - one or two team members - and then grew as we demonstrated more and more value.
What should you do next?
Psychologist Dr. BJ Fogg of Stanford University has created a simple yet effective approach to habit forming, and I think it’s worth a look to see what you can take away and apply to your own situation when trying to get sponsors bought into data quality. The key is to start very, very small. The focus is far more on getting started and completing small increments as opposed to launching your entire vision.
Tip: Here are a whole bunch of additional habit-forming tips based on academic research that you may find useful.
For example, instead of telling a data owner they are now responsible for 30-40 data items, why not start with five or even one, just for a month. Get them used to the habit first. Teach them how to master the technique of stewardship and governance for one data item. Which systems do they need to update, which users need to be notified, what change control processes need to be in place? There is a lot of plumbing and process involved just with the straightforward task of ownership, so don’t overwhelm them. Give them one data item to manage and then organically scale it up.
Next, think about your data stewards. Do you really want them to create hundreds and hundreds of data quality rules spanning 10+ dimensions? Why not give them one dimension to start with: completeness. Get them to manage completeness on one information chain in their domain. Once they're comfortable with that process, see if you can add more dimensions or rules. Then add more chains and data items under their control.
Instead of creating a data quality policy document that runs to 50 pages long, what are some simple start-up policies people can adopt with minimal fuss and investment?
Where can you apply this tactic?
There are so many areas of your organisation that can benefit from these small, incremental data quality habits. Where do you think your organisation could benefit? Please let me know in the comments below where I’ll be only too happy to share other tips on this valuable technique.
1 Comment
Great post. Starting small is so important.
I just wrote a post this week on getting users to care about data quality. I took the psychological perspective as well.
http://www.springcreeksystems.com/2013/07/01/users-care-about-data-quality/
In the end, we're all human and will respond to techniques based in psychology. Starting small & using positive reinforcement will improve data quality far more than trying to make everything perfect and punishing those who don't comply.