Quality and Innovation

Who’s accountable for data integrity? Most organizations don’t call it out directly, making the environment ripe for power differentials and blame. Chad Sanderson, CEO of Gable.ai, talked about it on Linked earlier today:

Too often, leadership teams lean on the Chief Data Officer and data/analytics teams to “fix the data” – source and process it from across all functional areas of a company and make the results useful.

I even saw a post yesterday from a senior leader accusing Data Scientists (all of us!) of “being lazy” for pushing back, asking to work with data that meets minimum standards. I’d like to ask this guy if he’d also be OK with:

• Living or working in a building that hadn’t been certified by structural engineers, and might not be safe for its occupants (those lazy engineers should just approve whatever they have to work with!)
• Eating at an establishment that hadn’t been approved by health and hygiene inspectors (those lazy restaurant owners just don’t want to cook!)
• Flying with an airline pilot who didn’t feel the need to do pre-flight checks (he’s not lazy… he’ll just fly whatever he’s got and the passengers can hope for the best!!)

Pushback doesn’t mean the pusher is lazy. It means they’re exercising their professional integrity – and that’s precisely the kind of data scientist you want.

When data teams are just order takers, working tickets to fix dashboards and figure out why a number on one report is different than a number on another report, 90%+ of the opportunities the organization has to improve data integrity are missed.

It’s also bad for morale, because every time a data team member hears someone say “the data is bad!” – they know they’re about to be held accountable for an issue that someone else is really in control of. (Accountability means getting real about what people can and can’t do.)

The best and most productive Chief Data Officer role I ever had had, at its core, a strong, equal partnership between the software and data teams and the divisions led by the Chief Operating Officer. in addition to answering research questions and building real time pipelines an visualizations, our data teams also were able to request that work be done by (today’s equivalent of) Product and Tech and Ops teams. If we saw that their definitions or concepts needed to be adjusted, or could see huge opportunities for improvement if sites or systems were adjusted, we could request that the work be considered and prioritized by those teams. Oftentimes, they appreciated the opportunity to hear about what they could do to align themselves with the rest of the organization… and the data teams knew exactly what those requirements for alignment were.

It’s not hard to establish the basis for good data leadership: just create an equal partnership.

A sign of GOOD LEADERSHIP is when the executive team forms an egalitarian partnership with the CDO, asking: “How can WE HELP YOU to help our ENTIRE ORGANIZATION fix its data and get better, more meaningful insights?”

In contrast, a lack of leadership is usually accompanied by demands on the data and analytics teams to “fix the data” or “make the dashboards more accurate.” All of those metrics depend on business processes that are owned by business teams and very likely cross organizational boundaries.

While a partnership between the business owners and the data team will easily fix the root cause of any observed issues, a power differential where the business team places an order and the data team has to work it… will not.

[I talk a lot about data and power differentials in my new book, Data, Strategy, Culture & Power. Check it out!]