Leading Analytics Podcast

The episode offers a handful of genuinely useful ideas - the 50/50 internal-showcase model, the three-bucket scorecard, and the privacy-risk compounding point - but large stretches are filled with generic leadership advice (employee empowerment, culture, change management) and throat-clearing. The actionable-ideas-per-minute ratio is low for a 32-minute conversation.

“our data strategy became to create an AI enterprise within IBM itself because we look a lot like our major customers and then use that as a showcase for our clients and customers”

“if you're in a data first culture or data driven culture and you have to go up for permission, then you're not in a data driven culture”

Most of the content recycles well-worn CDO orthodoxy - C-suite buy-in, business outcomes, change management, culture. The 'change agent in chief' framing is genuinely crisp and the startup-monitoring tip is somewhat counterintuitive, but these are isolated moments in an otherwise conventional narrative.

“he uses two words to describe it, which is basically change agent. He thinks of me as change agent in Chief”

“be very aware of what's happening in the startup space...It's a counterintuitive tip because uh, in most cases the startups and those smaller companies will be screened by people who are uh, playing different roles”

Bhandari is a genuine four-time CDO with six-plus years running data strategy at one of the world's largest enterprises - a real practitioner, not a thought-leader-for-hire. His credibility is high, though the conversation doesn't fully exploit the depth his tenure should allow.

“It's my fourth go around as Chief Data Officer and by far the most complicated”

“when I started out as chief, ah, data officer, that was 2006, um, there were only four of us globally with that title”

There are a handful of concrete data points (CDO community growing from 20 to 1,000+, four CDOs globally in 2006, monthly boss reviews, three strategy shifts in six years) but the episode conspicuously avoids naming a single specific AI use case, dollar figure, or measured outcome. The scorecard dimension literally uses 'X percent' as a placeholder.

“we had maybe about uh, 20 or so uh, data officers that uh, were part of that community. And uh, now it's well over 1000”

“reducing end to end cycle time by X percent”

The host asks broad, pre-planned questions ('tell us about your role,' 'what are the four or five qualities,' 'advice for aspiring professionals') and never challenges a vague or incomplete answer. The closing 'Did I kind of get that summary right?' exemplifies a PR-style chat rather than a probing interview.

“Did I, did I kind of get that summary, right, of our conversation?”

“Okay, talk to me a little bit about how you define success”