The Leadership Metric That Changed His Data Team’s Reputation with Jeff Baird

‍

When Jeff Baird talks about leading a data team, the conversation starts with trust — and one simple metric that transformed his team’s reputation: Days Since Last Incident.

On this episode of Crunching Data, Jeff shares how a 20-year career spanning scrappy startups to global enterprises taught him that people, not tools, are the ultimate driver of success.

Trust is the foundation

When Jeff joined a company where data pipelines failed almost weekly, dashboards were unreliable, and executive confidence was low, he knew technical fixes alone wouldn’t rebuild trust.

He introduced a metric anyone could understand: if a dashboard was wrong or a pipeline broke, the counter reset to zero. What started as barely hitting seven days grew to more than 200 consecutive days without an incident. The result wasn’t just stability, but a renewed confidence in the team’s work and a shared, visible measure of success across the company.

Hiring for impact, not headcount

Jeff hires using the “hungry, humble, smart” framework from The Ideal Team Player and is willing to pay top-of-market for the right people. The ROI can be enormous: when great engineers are trusted to do their best work, they can deliver millions in value.

Leadership is relationship work

From winning over skeptical stakeholders to re-hiring trusted colleagues multiple times, Jeff sees relationship-building as a data leader’s most undervalued skill. It’s also the first thing new leaders should prioritize.

For more on setting yourself up for success in those early months, check out The Data Leader’s Ramp-Up Guide.

The tech still matters, but it’s not the point

Jeff’s take on the Snowflake vs. Databricks debate is refreshingly pragmatic: there’s no “best tool,” only the right one for the business. The same philosophy applies to the ongoing unification vs. best-of-breed discussion. Balance is the goal, and business needs should decide.

Data modeling isn’t dead

Jeff’s final “spicy take” pushes back on the idea that modern tools make data modeling obsolete. While he isn’t dogmatic about star schemas or strict Kimbell methods, he believes the principles still matter — especially when preparing data for AI or natural language querying. Clear naming conventions, logical structure, and an understanding of data grain make analytics and AI results more accurate and more trustworthy.

Watch the Podcast above or here

New episodes drop bi-weekly. Subscribe to our newsletter Sync or Swim to make sure you're getting them.

Unitl next time.