To measure your podcast stats, you need to look at three sources: your hosting dashboard (downloads per episode), the dashboards of each platform where people listen or watch (retention, followers, demographics), and qualitative signals (comments, messages, mentions). None of them tells the whole story on its own. The key isn’t piling up numbers, but choosing a few that help you decide what to do in your next episode.
First, why podcast numbers are confusing
Unlike a video or a social post, podcast audio is distributed through an RSS feed that lots of apps read in their own way. That means:
- A “download” doesn’t always mean someone listened.
- Each platform measures different things under similar names.
- Your hosting numbers and an app’s numbers don’t match.
Nobody’s lying to you: they just measure differently. If you want to better understand how your podcast travels, read what a podcast RSS feed is.
The metrics that actually matter
Downloads or plays per episode
They tell you how many people are interested in each episode. The most useful approach is to compare them over a fixed window: for example, each episode’s downloads in its first 7 or 30 days. That way you compare episodes on equal terms.
Retention or listening time
Several audio and video platforms show how much of the episode people listen to and where they drop off. It’s probably the most useful metric for improving your content:
- If lots of people leave in the first few minutes, review your intro.
- If there’s a drop at a specific point, listen to that moment: it could be an uninteresting topic, a long pause, or an audio problem.
- If retention stays high all the way through, that format works.
Followers or subscribers
They show how many people decided to stick around. They grow more slowly than downloads, but they’re more stable.
Location and device
Where people listen to you (country, city) and on what (app, browser, TV). If you find out half your audience is in another country, you might want to adjust your schedule, references, or guests.
Qualitative signals
- Comments on platforms and social media.
- Direct messages and emails.
- Reviews and ratings.
- People who mention or recommend you.
A message saying “this episode changed how I do X” is worth more than any chart.
Where to find each piece of data
| Source | What it gives you | What you use it for |
|---|---|---|
| Hosting dashboard | Downloads per episode, per app, per country | Overall trend and comparing episodes |
| Audio platform dashboards | Followers, retention, demographics (depending on the platform) | Understanding how people listen |
| Video platform dashboards | Views, watch time, retention, traffic sources | Improving titles, thumbnails, and structure |
| Social media | Clip reach, comments | Seeing which topics spark conversation |
| Your email or forms | Messages, proposals, questions | Qualitative signals |
Each platform changes the names and location of its metrics fairly often, so check their help centers when you can’t find something.
How to build your dashboard without losing your mind
You don’t need special software. A spreadsheet with one row per episode and these columns is enough:
- Publish date.
- Topic and format (solo, interview, panel).
- Length.
- Downloads at 7 days.
- Downloads at 30 days.
- Average retention (if the platform provides it).
- Comments or messages received.
- A note: what you learned.
Review it once a month. Looking at the numbers every day only gives you anxiety.
How to interpret what you see
Some questions worth asking yourself:
- What do your best-performing episodes have in common? Topic, guest, length, title.
- Is there a publishing day that performs better? If you change it, change only one thing at a time.
- Does retention always drop at the same minute? It might be your structure.
- Are you growing or holding steady? Steady with good retention isn’t bad: you have a loyal audience.
Be careful about drawing conclusions from a single episode. A spike could be because a guest shared it a lot, not because the topic was better.
Vanity metrics vs. useful metrics
| Vanity | Useful |
|---|---|
| All-time total downloads | Downloads per episode over a fixed window |
| Followers across all networks added up | Podcast followers and their trend |
| Likes on a viral clip | People who reached the full episode from that clip |
| A one-off chart ranking | Retention and comments |
What to do with the data
Measuring is only useful if you change something. Examples of decisions:
- Retention drops at minute 3 → shorten the intro and get to the topic sooner.
- Interview episodes outperform solo ones → increase their share.
- One topic generates lots of messages → do a part two.
- Clips in a certain style bring people to the episode → make more like that. Vertical clips from your podcast will help.
- Your audience is in another country → adjust your schedule and references.
And if your goal is frequency, cross-check your data with how often to publish a podcast.
What you can’t measure with numbers
The most valuable part of a show doesn’t always show up on a dashboard: your relationship with listeners, the quality of the conversations, the network you build with your guests. Every so often, ask your audience directly what they like and what’s missing. The answers are usually clearer than any chart.
If, beyond the numbers, you want to improve how your show looks on video, create your studio for free and try out a new set.
Frequently asked questions
Why don’t my hosting downloads match Spotify’s numbers?
Because they measure different things. Your host counts downloads from every app that reads your feed, and each platform has its own way of counting plays. It’s normal for them not to match.
How many downloads are good for a podcast?
There’s no universal number. It depends on your niche, your language, and how long you’ve been at it. It’s more useful to compare your episodes with each other and see whether the trend is improving.
How often should I check my stats?
Once a month is enough for most people. Checking daily creates anxiety and leads you to change things without enough data.