What I learned analyzing 3 years of Huberman Lab Podcasts

And several unanswered questions…

GaryGeo
ILLUMINATION
5 min readMar 23, 2024

--

This article is a summary of analyzing the roughly 3 years of the Huberman Lab Podcast… what I learned and what open questions remain. This analysis focused on the trends and patterns of the published episodes to better understand the variety of episodes and how I might navigate the large collection to find the material I am interested in.

My analysis was based on YouTube details from 181 episodes from January 23, 2020 to February 1, 2024. The YouTube details included publication date, views and likes, in addition, I categorized the episodes and noted if they were a lecture, AMA or guest focused episode. I ended up writing an article for roughly each year (2021, 2022, and 2023) and this article is the compilation of those key insights.

What I Learned

#1 — The Huberman Lab Podcast videos were a hit from the first season, driven by Dr Huberman previous Instagram popularity and his guest appearances on popular podcasts like Joe Rogan and Lex Fridman.

This is a line graph representing the chronological view trends of 181 Huberman Lab Episodes. The x-axis is the date and the y-axis is the view count. There is a high degree of variability.
View trends over time showing high variability — chart by author

# 2 — The podcast started out with a lecture format, but in June 2021 guest interviews started to be incorporated (about every other week) and in 2022 Ask Me Anything(AMA) videos were added in 2023.

This is a stacked area graph representing the types (Lecture, Guest, etc) over time. Demonstrating that the first episodes were all Lecture and then a Guest formation was introduced. Recently AMA started appearing.
Episode Type Trend — chart by author

#3 — The Pareto Principle is at work with about 20% (30) of the 181 episodes having more than 2 million views and the majority having less, including a large number in the 1 million range.

20% of episodes account for over 50% of the views.

A 2 y-axis line chart. The left axis is the sum of views, the right axis is the count of episodes and the x axis is the view group (a bucket of the view range). This illustrates that about 20% of the episodes are responsible for 50% of the views.
About 20% of Episodes create over 50% of the views — chart by author

We can further illustrate this by comparing the top 3 episodes to the rest.

A scatter chart with all 181 episodes comparing the view count and like ratio. This is illustrating that there are a few stand out episodes with view counts that exceed the rest of the episodes.
Scatter chart comparing View Count with Like Ratio — chart by author

Understanding these trends have given me a sense of the production cycle and to better position myself to navigate the vast library of episodes by format, category and popularity. There were however observations that I don’t have an answer for.

Unanswered Observations

In pouring over this date and visualizing it multiple ways there were a few patterns that appeared which were not easily explainable. I can only speculate on what the reasons are as only people with direct experience with the podcast and/or YouTube might have the answers.

#1 — In 2023 the rate of published videos increased with multiple episodes a week. I graphed the gap between episodes and the precision is impressive. There has not been a span of more than 7 days between episodes and the cadence is increasing. What drove this? Part of the answer is publishing AMA and Live events, but that alone does not explain this trend, what were the other factors? Was the supply of content increased? Or maybe the Huberman Lab team has increased in size?

A line chart illustrating the cadence of new episode videos. It is very consistent for the first 2 years and then becomes variable as AMA and other episode formats are introduced.
Chart of days between episode videos being published — chart by author

#2 — There generally is an inverse relationship with the views and the likes (as a percentage). So that is to say that the more viewed a video is the like count as a ratio goes down. I suspect this could be related to the fact that one person can view a video multiple times, but like it only once… Or maybe more popular videos are promoted to YouTube users that are not part of the main fan base, making them less likely to leave a like?

A line chart demonstrating the general trend of episodes with higher views having a lower ratio of video likes.
The trend of episodes with higher views having less video likes — chart by author

#3 — The formula for a “Hit” episode appears to be tapping into a popular topic or celebrity guests. It would be interesting to better understand the programming strategy here. Is it clear to the Huberman Lab team that the content of some episodes will receive lukewarm views? If so, is it more important to have consistent frequent episodes to fill in the gaps, than just invest time in the “hit” episodes? Or maybe even the average episodes are a great investment in returns… or maybe it is just aligned with Dr. Huberman’s mission?

Below are the top and bottom 10 episodes to provide a sense of the extreme polar ends of the popularity distribution. The lowest episodes are Ask Me Anything and generally tend to be more recent (indicating views have not built up), but there are also many middle ground episodes. The hits have mainstream topics and celebrity guests.

A data table with the top and bottom episodes by views —table by author

If you have any insights into these trends and behind the scenes thinking please comment as I would love to know.

Wrap Up

As I wrap up this series of articles I am coming away with a deeper understanding of the Huberman Lab Podcast as a whole. What are the different categories and formats, what are the hits, and what is evolving over time. While not your typical, “What did I learn in episode x?” My intent is that by understanding the broader body of work, viewers/listeners will be better able to navigate and gain insight from this amazing resource.

Things to Know about this Data

  • My analysis is based on YouTube data as of ~February 10th, 2024. I do not have historical records of how the counts appeared at a set point in time (e.g. 1 month after airing), that might give a more equivalent evaluation.
  • “Liked Ratio” is based on the number of likes divided by the number of views. I believe a single person can view an episode multiple times, but cannot like it multiple times.
  • I used ChatGPT to summarize the episodes and provide the “Assigned Category” to create a grouping.
  • The data I based my analysis on is here.

--

--

GaryGeo
ILLUMINATION

student of life, steward of ideas, data geek, maker and product guy