7
7 Comments

Doing a podcast listener survey

We want to send out a survey to listeners to figure out what they like, don't like, and to better inform future episodes. Have any of you done that? How'd you execute it, what did you learn? Anything to keep in mind for those of us doing our first listener surveys?

on July 29, 2019
  1. 2

    I've wanted to do something like this for a while, but haven't gotten around to it yet. The closest I've come was this Twitter poll.

    One thing I've been doing as an alternative in the meantime is finally diving into listenership stats provided by podcast players. Spotify in particular makes it easy to see how far people listen through each episode, where they drop off, etc. iTunes has similar features. Sure, it's not as qualitative or as immediately insightful as surveys, but with a little work, you can learn some things.

    My process is that I'll find episodes where very few people dropped off, listen back to them, and take notes on what I thought was good or distinctive about the episode. I do the same with "bad" episodes where people dropped off very quickly. Then I compare all my results and try to extract some learnings. I've done this twice, and it took a few hours both times.

    Here are a few of the things I've learned:

    • Listeners consistently drop off at higher rates when the guest has an accent.
    • Solo founder episodes do much better than when I interview people who have teams.
    • Interviewing multiple guests on a single episode tends to perform poorly.
    • Listeners love episodes where experts come on and talk about their expertise, moreso than my typical founder interviews.
    • There's usually a sharp drop-off in the first minute of the episode. Cutting out my intro hasn't really changed that.
    • etc.

    I'll probably do more Twitter polls in the future, and maybe I'll eventually set up a survey for podcast listeners. But I do feel like I can make lots of improvements just from analyzing the stats I've got.

    1. 2

      To add to Courtland's points

      • Polling is only effective with really big audiences & even then it's still skewed, looking at the cold hard data will tell you more about consumption since it takes the silent majority into perspective & Courtland has a great strategy for interpretation
      • Pretty much all shows will see that steep dropoff up front so don't pay it too much mind, instead focus on what elements your best performing episodes had
      • Also, not every episode will perform the same. You never know what's going to pop or flop and there are so many circumstances at play to determine the popularity of any given show you can only really take the data so far. Even at a network level, we're still guessing demographics, listener habits, and forecasting impressions for ad sales
      1. 1

        Thanks, good insight!

    2. 1

      I did not realize that I could get these kinds of stats from Apple Podcasts and Spotify. Thanks for highlighting! I'll definitely check those out.

      I noticed that you use BackTracks for podcast hosting. I was considering using them because they appeared to offer advanced analytics, but they were too expensive for me to justify for a hobby project and I could not tell how they interacted with analytics from the podcast directories themselves.

      @csallen Have the analytics from BackTracks been useful?

      1. 1

        They've been pretty generic, nothing mind-blowing. But I haven't used other hosts, so they're all I'm used to. Perhaps they are way better than alternatives.

    3. 1

      Some good insights @csallen. Do you think the accent thing is just accents in general or an accent that is hard to understand?

      1. 2

        My guess is that it's because people find accents hard to understand, but I've seen this effect happen with British, Irish, Nigerian, Indian, and Eastern European accents, and I personally didn't find many of these people difficult to understand at all. 🤷‍♀️