Analyzing Podcast Transcripts

Asking great questions generates great conversation, which in turn builds trust. Leaders that know this find that asking great questions is often a catalyst for truly thinking together. And thinking together can put you on the path to solving intractable problems and sparking innovative thinking. On Sanjay Gupta’s podcast, Celeste Headlee goes deeper: “We don't ask enough questions. Questions are so powerful at making other people feel heard. Not even necessarily your first question, but there's a special power to follow up questions that makes people feel that they are liked, that they are heard, and that you're listening.”

How do we learn how to use a question framework to create high-performing teams? One place to look is the Freakonomics podcast - a podcast dedicated to exploring the hidden side of everything. These explorative episodes require a lot of question asking and, judging by the number of listeners, embody captivating conversation. Can analyzing transcripts from several Freakonomics podcast episodes help us understand how to better leverage asking questions to create great conversation? Great conversation that builds trust and solves problems? Maybe.

As a proof of concept, analyzing only 7 episodes worth of transcripts does indeed give us insight. We can extract all the questions asked during an episode. On average, episodes are ~57 minutes and contain ~52 questions. This calculates to about 1 question per minute. Another way to convey this is that roughly 12% of an episode is question versus non-question sentences.

Once we have the questions extracted, a preliminary cluster analysis helps us roughly categorize questions into 3 types; what, why, and how questions. The below graphic simply visualizes how well we can separate the types of questions using mathematical techniques (dimensionality reduction + clustering).

Coincidentally, Brian Oshiro gives a great TED talk emphasizing the importance of these types of questions, saying they are all needed in education, but there is an overemphasis on what questions. If there is to be risk and uncertainty in the world (which, obviously, there inevitably will be), then you need more why/how questions to build critical thinking skills.

We can look at examples from each category. Judging from the graph above, I suspect the how and what questions won’t separate 100% cleanly…

What:

But can an elite degree really change your life?

As the Notre Dame sociologist Joel Mittleman put it “If America’s gay men … formed their own country, it would be the world’s most highly educated by far.” But fewer than five percent of men in the U.S. identify as gay, so for the rest of the young men who aren’t going to college, but might benefit from it —
what should be done?

And speaking of crime, did the “defund the police” movement really catch fire?

Why:

Why is it that these schools are the ones that haven’t expanded even though we see more students going to school?

Why are you trying to keep this sort of club and this sort of prestige?

Today on Freakonomics Radio
why are sports so useful in trying to burnish a reputation?

How:

And if so, what would that actually look like?

And
how is that different from any sort of reputation laundering?

So,
how do you try to find that fruitful medium?

We can also look at how these question types are used over the course of a podcast episode. Using a single episode as an example to visualize… the episode starts out with how & why questions, followed by a streak of what questions, and then mostly bounces back and forth between how & what questions.

Looking at multiple episodes, there’s no strict pattern. What is more interesting is that the use of why questions is quite minimal. There appears to be more preference for how & what questions… and the strongest preference for how questions. Across the 7 episodes analyzed, the majority category was how questions:

Original question: Can podcast transcripts help us understand how to better leverage asking questions to create conversation that builds trust and solves problems? This brief analysis hints to us that to create great, engaging conversations you want questions to be 1) frequent and 2) heavy on how/what questions rather than why questions. What do you think? Do these insights have you rethinking your question framework?

Be visionary.
Try Storied today.

You speak 3-4x faster than you type. Create writing at the speed of AI that truly resonates and drives action.