What Gets Measured

There are a handful of genuinely useful observations - dashboards designed to generate questions, AI amplifying existing shortcomings, the bifurcation of analytics roles - but they are spaced out by extended host rambling, a trivia game, and repeated throat-clearing, bringing the ideas-per-minute ratio to merely average.

“a dashboard by its very nature is designed to be unsatisfactory 60 to 70% of the time”

“AI has the ability to take every one of your shortcomings and escalate them at a higher velocity”

The political/interpersonal framing of data analytics and the 'commando analytics vs. governance' bifurcation offer modest novelty, but the core thesis ('business problems are people problems') is a well-worn consulting talking point, and most other claims recycle standard analytics conventional wisdom.

“data has become more accepted, it has actually become a vehicle of political discourse in a company, um, and a power source”

“Dashboards are the sense of continuity to say, back in fall of last year, we decided this is our strategy. How is that playing out?”

Shiv Gupta is a genuine 25-year practitioner who started as an econometrician and has operated across insurance, healthcare, and advertising at scale, lending real credibility; however, he functions primarily as a consultant and thought-leader rather than an in-seat operator who built and ran these functions inside a large organisation.

“I started off as an econometrician”

“we were working on a project where we were trying to estimate lifetime value for a, um, insurance product”

The insurance LTV-to-sales-language anecdote is the episode's only concrete example, and even it lacks named companies, timeframes, or hard numbers; the '30 billion in revenue impact' appears only in the host's intro bio and is never substantiated in conversation.

“you can go right into their book of business, take out 20 customers and say, according to lifetime value, this is going to be your best customer”

“The CFO doesn't care about your click through rate. The CFO doesn't care about which campaign had which conversion. It's irrelevant to them”

The host is enthusiastic and occasionally draws out a good example (the insurance LTV story), but questions are routinely long, self-answering, and structurally confused, and the episode closes with several minutes of a trivia game that contributes nothing; there is essentially no pushback or productive disagreement.

“You know, it's interesting because I've been, you, uh, know you got your marketing data analysis, you put it in your report and then you share it with people. And it's interesting because maybe we're looking at a recurring data set, but we're asking deep dive ad hoc questions which requires another set of analysis. How often do you see people, uh, kind m of miss attributing the level of analysis that's required based on those two buckets that you're saying?”

“Anyway, donuts or beignet”