How to read podcast analytics in the right order

Read podcast analytics in a fixed order, because the order decides what you conclude. Start with one episode's download curve, then compare it against your own recent episodes, and only then open the app and device breakdown. Most people start with the monthly total, which is the one view that cannot tell them anything actionable.

The monthly total moves for reasons that have nothing to do with your last episode. A back catalog surge, a month with five publish slots instead of four, a bot wave your host filtered late. It aggregates away every signal you might have acted on.

How to read podcast analytics: the five passes

Work through these in sequence. Each pass answers one question and hands the next pass a narrower one.

Pass 1: one episode, its first 30 days

Open the download curve for your newest episode and read its shape, not its total.

A normal curve rises steeply for two or three days, then bends and keeps climbing slowly. That bend is your followers finishing their automatic downloads. What happens after the bend is the part worth studying, because that is discovery, and it is the only part you can grow without publishing more.

Two shapes are worth reacting to. A curve that goes almost vertical then completely flat means you reached your followers and nobody else, which is a discovery problem. A curve with a second bump a week or two later means something external sent people to that episode, and it is worth finding out what.

Pass 2: the same episode against your own last four

Buzzsprout's episode pacing chart plots your four most recent episodes from day one to day 30 on one set of axes, which is the right comparison to make. Whatever host you use, build the equivalent view.

You are looking for whether the newest line sits above or below the other three at the same age. Comparing a three-day-old episode's total against a three-week-old episode's total tells you nothing except that time has passed, and it is the single most common misreading in podcast analytics.

Benchmarks come second, and they come with a date attached. Buzzsprout's platform stats for August 2026, covering 112,207 active podcasts, put the median episode at 27 downloads in its first seven days. The top 25 percent reach 96, the top 10 percent reach 407, and the top 1 percent reach 4,526. Our download benchmarks piece covers how to place yourself against figures like these without drawing the wrong conclusion from them.

Pass 3: the app and device split

Now look at where downloads came from. Buzzsprout's August 2026 breakdown gives a sense of the normal shape: Apple Podcasts at 35.8 percent, Spotify at 28.8 percent, web browsers at 7.2 percent, and Castbox at 1.9 percent. Mobile accounts for 85 percent of downloads, computers 8 percent, and smart speakers 0.6 percent.

Your own split will differ, and the absolute numbers matter far less than sudden movement in them. A platform that drops by half in a week is almost never an audience event. It is a feed problem, a prefix problem, or a change in how that platform counts. Our guide to downloads dropping works through those causes in likelihood order.

An unexplained rise in the web browser or unknown category deserves the same suspicion. It often means requests are arriving without a recognisable app signature, which is what unfiltered traffic looks like.

Pass 4: geography

Read geography once a month, not weekly, because it moves slowly. Buzzsprout's data has the US at 52 percent, the UK at 5.7 percent and Canada at 4.4 percent across its platform.

The use for this is commercial rather than editorial. A sponsor selling into one country cares what share of your audience is actually there, and a show with 40 percent of its audience outside the sponsor's market should price and pitch accordingly. It also tells you whether your release time makes sense for the timezone most of your audience lives in. Our listener demographics guide covers how to turn this into something a media buyer can use.

Pass 5: retention, if your platform reports it

Downloads stop being useful here, because they are recorded before anyone has pressed play. Apple and Spotify both report playback, and Spotify counts a play only after 30 seconds, which makes its figures closer to attention than Apple's.

What you want is the drop-off point. If a consistent share of listeners leaves at minute two, the problem is your open. If they leave at minute 25 on a 45-minute show, the problem is the middle. Our notes on improving listener retention cover what to change once you know where the leak is.

What to stop reading

All-time downloads. The number only goes up, so it always looks like growth. It flatters old episodes and tells you nothing about the last three months.

Yesterday's number. Apps sync on their own schedules, so daily figures bounce for mechanical reasons. A weekly figure at the same point in each episode's life is the smallest window that means anything.

Any single combined audience number. If your dashboard adds host downloads to platform plays, it has combined two counters with different thresholds and produced a figure you cannot defend to a sponsor. What counts as a podcast download covers the counting rules, and our podcast analytics guide covers which metric answers which question.

Run the five passes once a month and the first two weekly. The discipline is in the order, because a dashboard read from the top down will hand you a conclusion about your show that is really a conclusion about your publishing calendar.

Podder reports per-episode curves, app and device splits, and unique listeners from the prefix, so all five passes come from one source instead of four dashboards that disagree. Start on the Analytics plan and install the prefix on your next episode.