The world of podcast metrics can get confusing. We've got downloads, streams, plays, audience, listeners, views, watchers, community members, and maybe even more terms than that. But when you boil it all down, I think it comes to three basic labels: downloads, plays, and audience. Here's what each one actually means, and the caveats you need to know for each.
1. Downloads (and “streams”)
This is the classic measurement. For many years, the benchmark question was simply, “How many downloads are you getting?”
A “stream” is usually still a download. In most modern podcast apps, you can open an episode, press play, and start listening within a few seconds. That feels like streaming, but what's really happening is a progressive download. The file starts downloading the moment you press play and buffers ahead in the background. Because connections are fast, it looks instant. Some apps or data networks download in chunks (50 MB, 100 MB, and so on), so a long episode might download halfway first and finish the rest when you reach that spot. Other apps download the whole thing at once. Either way, the file is being downloaded from the internet onto your device. That's still a download.
Downloads are the most fundamental measurement of a podcast's reach because they're tracked at the server level. But they're reliable, not accurate. There are plenty of bots out there, both good and bad, that download episodes to transcribe, cache, or gather information automatically without a human ever pressing play. Hosting providers like Captivate, Buzzsprout, Blubrry, Libsyn, and Transistor maintain lists of what they consider bots and filter those out of your counts (while usually still serving the file so nothing breaks).
Even with perfect filtering, downloads don't tell the whole story. I've talked before about the myth of monthly downloads. Saying “we got 100,000 downloads this month” tells you almost nothing about your audience size. Here's an extreme example to illustrate the point: you could release 100,000 episodes all downloaded by one person, and your audience would be one. So downloads per day, week, or month is essentially a meaningless metric for understanding audience.
The one place downloads-per-month is useful is advertising. If you have dynamically or programmatically inserted ads across your whole catalog, downloads per period tells you how many ad placements could be served.
But if you want to understand your audience size, the metric to use is downloads per episode after 30 days. That's the classic benchmark used for advertising and for understanding your audience, because it captures both your most loyal listeners (who download in the first few days) and the stragglers. Just remember it only applies to an episode that's actually 30 days old, and you should compare episodes using the same amount of time since publication. Many analytics tools (OP3, Captivate, Transistor, Blubrry, Buzzsprout) now show various benchmark periods like 1, 3, 7, 10, 15, and 30 days, but all of them include 30 days because that's the long-standing benchmark.
This approach also handles the messiness of IP addresses. If you try to count “unique listeners” by counting unique IP addresses in a month, you'll get a number that's too big. One person can rack up multiple IP addresses for a single episode: they start streaming at home on Wi-Fi, get in the car where the cell IP changes, then pull through a coffee shop drive-through where the phone briefly hops onto that Wi-Fi. That's potentially three IP addresses for one episode. Across multiple episodes, one person might represent home, work, and mobile addresses. You can try dividing by the number of episodes released, but that often cuts the number down too aggressively.
Downloads per episode after 30 days sidesteps a lot of that, because once a person downloads an episode they probably won't download it again. And when an IP address changes mid-download, many hosting and analytics providers are smart enough to see the same user agent picking up right where the last chunk stopped and stitch it back together as one person.
The takeaway: downloads are very reliable but not very accurate. They require a lot of filtering and algorithms to account for every scenario, which is exactly why we need other ways to measure reach.
2. Plays
So what is a play? That's maybe the question Shakespeare would ask if he were a podcaster today. To play or not to play?
In places like Apple Podcasts, a play is counted when the episode is played. That does not mean it was listened to completely or finished. It just means it was played. That's why you might look at your stats and see your plays are actually much higher than your downloads or your follower count. It counts every play. If someone pauses and presses play again, that's another play. If a voice assistant pauses playback to give driving directions and then resumes, that can be another play. Drive across town with turn-by-turn directions interrupting your episode and you might single-handedly account for dozens of plays.
So plays are interesting but not very reliable. That's where the Alliance for Measurement in Podcasting (AMP) comes in. They're a group that's been working behind the scenes, like a man behind the curtain, trying to define what a play actually is. I don't love how secretive they're being or who they're leaving out of the group, but I do like what they're trying to do. If we could all agree on what counts as a play, that would be genuinely beneficial, especially if it could extend to non-podcast platforms like YouTube.
For comparison: on YouTube it's assumed (never confirmed) that a view isn't counted until at least 30 seconds have played. In podcasting, the IAB (Interactive Advertising Bureau) download guidelines say a download counts once 60 seconds of actual audio has been downloaded, measured after header information like chapters, ID3 metadata, and the embedded image. AMP is trying to define a play as 30 seconds, and they may also define whether resuming after a pause counts as a new play.
The core challenge is where each thing is measured. Downloads happen at the server, so hosting providers and third-party analytics can see them. Plays happen inside the app, where those services can't see what's going on. This is also work that could fit the Podcast Standards Project (PSP), and it's why I've long wanted a standard way of measuring things. When we use a term, the majority of platforms and apps should understand it to mean essentially the same thing. The IAB download guidelines tried to do that for downloads, but there's still wiggle room and ways it can be faked.
Special thanks
- Ralph Estep Jr. invited me onto Podcasting Morning Show to talk about podcast engagement, chapters, and more podcasting topics.
- Martin Lindeskog streamed 706§ and boosted 1,111§ saying, “I broke your ‘rule' with my first podcast, EGO NetCast. It is caricature of yours truly, created by editorial cartoonist and fine artist, John Cox. I am wearing a baseball cap, holding a retro microphone and a radio. Thoughtful advice about not wearing headphones. I have found a pair of very comfortable headphones, AKG K92.”
3. Audience
Some people call this “listeners,” but I prefer audience. Podcasting has always included video, so “listeners” implies audio only. “Audience” covers both your listening audience and your viewing audience.
This is the third major way to measure a podcast's reach: how many actual people are consuming your show. And unfortunately, there's no universal way to measure it everywhere. There are real privacy concerns. If a podcast app can track what you listen to via some ID tied to you, it could build a picture of your consumption and sell that to advertisers. Some app developers have flatly refused to do anything that identifies a person or reports the number of consumers, and audience has to be tracked inside the app to begin with.
Despite the privacy concerns, there are ways this could be done. One idea, discussed by Dave Jones behind Podcasting 2.0, is the Universal Listener ID or Universal Audience ID (ULID/UAID). The concept: a podcast app generates a random string of letters and numbers assigned to a person and sends it with each download request. No matter what IP address that person is on, their downloads get tracked under that one ID.
But this raises privacy concerns of its own. If an app uses the same ID across multiple podcasts, analytics tools could start seeing everything you listen to. Apps could randomize it so you get a unique, globally unique ID per podcast. Yet even then, a malicious party watching the server could track all the IP addresses tied to one ID and build a picture of where you go. I wouldn't do that, and most hosting providers wouldn't either, but advertisers would love it.
This is the same reality behind “my phone is listening to me.” The truth is scarier: your phone isn't listening, the algorithm is watching. It notices that you and another person frequently visit the same place around the same time and share social connections, so it infers you probably talk. That kind of profiling is far more invasive than eavesdropping, and it works by building a profile of someone, often without meaningful consent. On social apps you technically consent through terms of service that everyone just clicks “agree” on. Podcast apps don't want to require that kind of consent, though Apple Podcasts does generate a unique ID for you (one you can reset anytime on your iPhone or Mac, though almost no one does).
There's also the handoff risk seen with browser extensions and plugins: you agree to one privacy policy, the developer sells the project, and the new owner changes what's tracked. It's happened with Chrome extensions, Firefox extensions, and WordPress plugins. A ULID/UAID could absolutely be implemented to respect privacy, but only if every app developer commits to it and those commitments hold.
The most important metric is the one that measures your goals
The reason I made this episode is to help you understand the caveats. I'm a bit of a Mr. Caveat. But once you understand how downloads, plays, and audience are each measured, you can read any platform's numbers with the right context instead of chasing the latest metric or standard.
For example, in Apple Podcasts you can see detailed consumption data because they track it inside their ecosystem. Knowing that plays counts every play, you'll understand why a long episode racks up more plays (people pause and resume long episodes more), and that a higher play count does not mean a bigger audience. Plays does not equal audience. Downloads is closer to equaling audience, but even a download doesn't guarantee someone actually consumed the episode. Edison Research has shared survey data suggesting most people who download an episode listen to most or all of it, but that may be shifting as more “downloads” are really progressive streams, and many listeners don't know the difference.
So it really comes back to what matters for you. Some of these numbers genuinely matter if you have sponsors, advertisers, or a business that needs KPIs. There's a real need for that in those cases. But for many podcasters, the fuel to keep going isn't the size of the audience, it's the depth of your relationship with the audience. The more you foster that relationship, the more that becomes the number that matters.
How many people emailed you this week? How many commented on your episode? And here's something cool I show inside Podgagement: not just your total reviews, but the ratio of ratings to reviews. In Apple Podcasts you can leave a rating without a review, but every review includes a rating. A smaller gap between ratings and reviews signals a more engaged audience. For The Audacity to Podcast, that gap is much smaller than for a show like Serial, which has a huge number of ratings but a big gap before reviews. If podcast reviews matter to you, collect and track them with Podgagement!
Of course, even reviews are one per person, so maybe the metric that matters most for you is feedback, community engagement, or social reposts. Think about why you're doing your podcast in the first place. Think about your P.R.O.F.I.T.: popularity, relationships, opportunities, fun, income, or tangibles. Which metric measures the why for your show? Is it audience size, plays, downloads, or something else entirely? If fun is your goal, then maybe the only metric you need is a simple thumbs up after each episode: did we have fun? If yes, you met your goal, and it doesn't matter how many downloads, plays, or audience you had.
It's up to you to decide what's important for your show. The most important metric is always the one that measures your goals.
If you love The Audacity to Podcast and value the podcasting inspiration and education I provide, would you please consider giving back what it's worth to you?
Supercharge your podcast engagement and grow your podcast!
Do you ever feel like your podcast is stuck? Like you're pouring your heart into your podcast but it seems like no one is listening?
Try Podgagement to help you supercharge your podcast endgagement!
Get speakable pages to simplify engaging with your audience, accept voicemail feedback (with automatic transcripts), see and share your ratings and reviews from nearly 200 places, follow your podcast rankings across nearly 34,000 global charts, discover networking opportunities, and more!
Ask your questions or share your feedback
- Comment on the episode
- Send a written or voicemail message here
Follow The Audacity to Podcast
- Apple Podcasts, Spotify, other Android apps, or in your favorite podcast app.
- Subscribe on YouTube for Podcasting Videos by The Audacity to Podcast
- Follow @theDanielJLewis on X-Twitter
Disclosure
This post may contain links to products or services with which I have an affiliate relationship. I may receive compensation from your actions through such links. However, I don't let that corrupt my perspective and I don't recommend only affiliates.
Related Content
