Every podcaster we talk to has had the same experience. You publish an episode, the download number goes up, and you feel good about it. Then the next episode does half the downloads and you feel terrible. Neither number told you anything about whether people actually listened.
Downloads measure delivery. They count when a file is fetched — by a person, by a podcast app refreshing a feed in the background, by a script. One listener with three devices subscribed to your show can produce three downloads of an episode they never press play on.
What actually happened to your last episode
The four numbers we track in PodAnalyst are unique listeners, listening hours, plays, and followers. Of those, listening hours is the one that most reliably tells you whether an episode worked, because it can only go up when someone stays.
| metric | what it measures | what it's good for |
|---|---|---|
| Downloads | Files delivered | Feed health, syndication problems |
| Plays | Episodes started | How many people tried it |
| Unique listeners | Distinct people who listened | Audience size |
| Listening hours | Total time spent listening | Whether the episode held people |
A download spike with flat listening hours is a syndication event — a new app picked up your feed, or a directory re-crawled it. Interesting, but it isn't growth. A small download number with rising listening hours is the opposite: fewer new people, but the ones you have are staying longer. That's the one to pay attention to.
Downloads tell you the file arrived. Listening hours tell you the person stayed. Only one of those is your show doing its job.
Reading a retention curve
A retention curve is a line that starts at 100% when the episode begins and traces how much of your audience is still there as it plays. In PodAnalyst it lives on the episode detail page, and it's the single most useful chart we render.
Here's how to read one without fooling yourself:
- Ignore the first 30 seconds. Apps buffer, people skip intros, and some plays are accidental. The curve always dips there. It is not a signal.
- Find the biggest single drop. That's your edit point. Something happened there — an ad break that ran long, a guest who changed the subject, a segment that didn't earn its minutes.
- Look at where the curve flattens. The flat part is where your show is working. The length of the flat section is, roughly, how much of your episode the audience considers mandatory.
- Compare across episodes, not against an absolute. A 40% completion rate means one thing for a 90-minute interview show and another for a 12-minute daily.
The drop that isn't a problem
Not every cliff is a failure. A hard drop at the end of your episode is usually people finishing and closing the app — that's a perfectly shaped curve. What you're looking for is a drop in the middle, where the episode had more to give and the audience disagreed.
Exporting the numbers and doing it yourself
If you want to look at this outside PodAnalyst, Pro and Network plans export your episode data as CSV. The retention column is the one worth plotting:
episode,minutes_elapsed,listeners_pct
ep-142,0,100
ep-142,5,78
ep-142,12,61
ep-142,31,58
ep-142,44,34
Row by row, that's a show that loses its audience early, earns them back by minute 12, holds steady through a long middle, and loses a third of the room at minute 44. That's a specific, fixable thing. "We got 4,000 downloads" is not.
PodAnalyst pulls this data straight from Apple Podcasts and Spotify, so the curve you're reading is the platform's own first-party measurement — not an estimate from a sample. You can connect your show in about five minutes, or read more about how retention is calculated first.
If you want to see it on your own numbers, start free — the Community plan includes retention data for one show.