What is a retention curve, and how do I read one?

If you only ever look at one chart about your show, make it this one.

A retention curve is a line showing how much of your audience was still listening at each moment of an episode. It starts at 100% and, in almost every case, it slopes down. Where it slopes, how steeply, and where it flattens out is the closest thing podcasting has to a direct answer to "did this episode work?"

It's also the chart people most often misread. So let's start with what it's actually drawing.

What the two axes really plot

Neither axis is the one people assume:

The first row trips people up constantly. A retention curve doesn't know what day it is. Every episode gets its own curve, and that curve describes one listening session, averaged across everyone who pressed play.

The second row matters more. Because the line starts at 100%, it's relative. The same curve can belong to an episode with 200 listeners and one with 20,000. Retention tells you how well the episode held the room. The listener count tells you how big the room was. Those are different questions and you need both answers.

Retention is a percentage. Audience is a count. A show can grow its audience while losing the room — and the reverse happens just as often.

In PodAnalyst the curve comes from the platforms' own first-party measurement — Apple Podcasts, Spotify, and YouTube each report it — so you're reading what they observed, not a model's estimate. Episode detail covers where each platform's numbers come from and how they disagree.

Reading one: four passes

Don't stare at the whole line at once; it's too much information. Read it in four passes, in this order.

1. Ignore the first 30 seconds. The curve always falls off a cliff at the start. Apps buffer, intros run long, and some plays are accidental taps. Every show on earth has this cliff. It isn't a signal and there's nothing to fix.

2. Find the biggest drop after that. This is the important one. Scan from left to right and find the steepest fall that isn't the opening. That timestamp is the most valuable number on the page, because something specific happened right there — an ad break that ran long, a guest who changed the subject, a tangent that didn't earn its minutes. Drops have causes. Find the cause.

3. Find where the line goes flat. The flat stretch is where your show is working — people are settled in and staying. Roughly speaking, its length is how much of your episode the audience considers mandatory. If that flat stretch is 15 minutes into a 60-minute show, you learned something about your real format.

4. Look at the tail. A steep drop at the very end is usually people finishing and closing the app. That's a well-shaped curve, not a failure. What you're hunting for is a cliff in the middle, where the episode still had more to give and the audience disagreed.

The shapes, and what they usually mean

A worked example

Here's a curve from a 52-minute interview episode. Read the table the way you'd read the chart:

Pass by pass: the opening cliff costs 18 points in the first minute, which is normal and not worth touching. From there the line is almost flat through minute 22 — so the first 22 minutes are effectively mandatory listening. Then between minute 22 and 23 it loses 17 points in a single minute. That's a cliff, and it has a cause. After it, the curve settles again and holds to a soft landing at the end.

The read: this episode is a 22-minute show with a 30-minute appendix. The single most useful thing the producer can do is open their own notes at minute 22 and find out what happened. Was there a break? A new segment? A guest swap? Whatever it was, it's the one edit worth making.

That's the whole method. A specific, fixable thing beats "we got 4,000 downloads" every time.

Four ways a retention curve lies to you

Comparing percentages across different lengths. 40% completion means one thing for a 90-minute interview and something else entirely for a 12-minute daily. Compare an episode to other episodes of the same shape, never to an absolute number you picked up somewhere.

Judging on a single episode. Curves wobble. One episode's cliff can be a bad guest, or it can be a Tuesday. Look for the same drop in two or three episodes before you change how you make the show.

Reading a line that mixes platforms. Every platform is plotted on the same chart by default, and their audiences behave differently — Spotify listeners sample more widely, Apple listeners are more committed, YouTube viewers tend to watch further in before leaving. A gap between the lines is expected. It's when the shape differs that you've found something platform-specific. Use the platform filter to isolate one at a time.

Forgetting the window. An episode does almost all of its listening in its first two months, which is why episode detail defaults to a 60-day view — it keeps the comparison like-for-like. If you switch to all time, you're comparing a two-year-old episode's full life against one that came out last week.

What to do with it

One thing. The largest mid-episode drop becomes your only production note for the next episode. Not five notes — one. That's the habit that compounds.

If a drop lines up with something you did on purpose, you can pin a note to that exact point on the chart in PodAnalyst, so the explanation lives with the data instead of in your head. And if you'd rather just ask, Iris will read the same curve for you — try "what happened at the drop in this episode?"

For the wider habit this fits into, see the 15-minute weekly review. And if you're still deciding which numbers to trust in the first place, why downloads lie is the place to start.

---

PodAnalyst pulls retention straight from Apple Podcasts, Spotify, and YouTube, 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.

If you want to see it on your own numbers, start free — the Community plan includes retention data for one show.