Podcast Stats are Weird—Here's How to Read Them
Confused about your podcast metrics?
Join the club. The way podcast stats are displayed, talked about, and used are all confusing. Add to that the general anxiety that can come from looking at anything that resembles a spreadsheet or graph, and you’ve got yourself a recipe for a mess.
In this article, I’ll walk you through some of the key basics, demystify some commonly misunderstood metrics, and help you make sense of what it all means (including if it means anything at all).
Now, I’ll be upfront with you. This article is one those “I’ve had this talk 17 times, I really should write it down” kind of articles. The thing about this particular talk is that I have hunch that people don’t believe me. So I’m hoping that putting it in writing—with charts!—will lend some credibility to my words.
First, why are you curious about your podcast stats?
Podcasters tend to get curious about their stats for a few reasons: they’re trying to respond to a potential sponsor, they’re trying to justify the time or expense, or they’re trying to figure out what they’re doing right (and wrong). So they—maybe you—crack open the analytics tab in their podcast dashboard and take a look.
And what they see just doesn’t make much sense.
That’s okay because: data doesn’t make sense on its own.
You can’t look at a dashboard of numbers and expect a story to emerge. Instead, always approach any set of metrics with a question in mind. Take the reason you’re curious about your podcast stats and turn it into a specific query.
What do you want to know?
How much I can charge for advertising on my podcast?
Is my podcast growing, or is it losing subscribers?
What episodes have performed well?
What episodes are people interested in from the back catalog?
What apps are listeners using?
You can answer all of those questions (to varying degrees) using your podcast metrics. And, by starting with the question, you’ll know what to look for.
Your podcast stats can’t tell you whether or not you’re a good podcaster (or a good human being). If you’re looking for validation, I guarantee you that your podcast metrics can’t give that to you.
Your podcast stats probably can’t tell you whether making your show is worth your time. That depends on why you started the show and what you hope to achieve with it. I’ll talk more about this toward the end of the article.
Next, let’s get some basics out of the way.
How do you know how many subscribers your podcast has?
Well, you can't. You can give it a good guess (I’ll get to that in a bit), but there is no definitive way to know how many people have subscribed or followed your podcast across all the different apps.
What is a download?
This is the least weird part of podcast stats. A download is exactly what it sounds like: a download of an episode. Do I mean a listen? No, I don’t mean a listen. Just a download. The file is downloaded and saved on someone’s device. It might also be streamed (and, yes, that’s probably a listen, but no, you can’t tell them apart from a regular download).
What counts as a download?
"Downloads" are any request for the file of the episode that your podcast host receives. That can be an automatic download—subscribers or followers opt to have new episodes downloaded whenever they're published. Or it can be a proactive download—a casual or new listener selects and downloads a particular episode. And, as I mentioned, it can also be a stream. In that case, a listener may select an episode and play it without downloading it to their device—just as you'd watch a video on YouTube or Netflix.
How many downloads is a good number to aim for?
It depends on your goals. If you want to seriously consider outside advertising, you'll probably need to show that new episodes receive at least 2000 downloads within the first week of airing. The number may be lower if you reach a small niche of hard-to-find customers. And the number will be considerably higher if your audience is more general (think 5-10k per episode, at least).
Most of the podcasters we work with don't care about outside advertising. They're self-supported through direct payments or self-sponsored through their own products or services. Then, the number of downloads is really just a function of the number of clients or premium subscribers you need. Often, the ratio is somewhere between 50:1 and 100:1 (average downloads per episode:conversions).
Wait. Where are my stats?
The stats for your podcast are connected to your podcast host. And yes, that's a confusing name because you are also a podcast host, right? In this case, your podcast host is the software service you use to "host" the media files for your show and (most likely) the RSS feed that distribution platforms connect with. Examples of podcast hosts include Transistor.FM (my personal favorite), LibSyn, Blubrry, Anchor, SquareSpace (not recommended), Substack, etc.
Once you've logged into your podcast host, stats or analytics are often the first thing you see. If not, there'll be a link in the navigation to view your stats.
Okay, with those common questions out of the way, let's look at the hairier aspects of understanding podcast stats.
Why don’t podcast stats offer more information?
If you’re used to website analytics or advertising metrics, the amount of data that your podcast host can provide is appallingly limited.
You’ll never know how people found your podcast, what episodes they listened to, or what they did after they listened. These are commonly available metrics in other media. But podcast stats don’t deliver on them.
The reason we don’t get more information out of our podcast metrics is the technology they’re based on. The backbone of a podcast is an RSS feed. And podcast apps are glorified RSS readers. RSS feeds are designed to be minimal—and that means that they don’t support the scripts required for things like Google Analytics or the Facebook Pixel.
But that’s okay.
I will happily argue that having less information is good for you and your podcast.
It means you’re less likely to try to make what “everyone wants,” and, instead, make what you want to make. You can move away from the false clarity that comes from guzzling data points and nurture your creativity with more holistic forms of feedback.
A final note on the basics: statistical significance matters.
“Statistically signficant” simply means that a data point has changed in a meaningful way—that is, the meaning of the change is noteworthy. A statistically insignificant change means that the change is within the margin of error. It could simply be a fluke.
If your show receives fewer than, say, 200 downloads per episode, that’s fine! And it means that you have to take any short-term changes with a grain of salt.
Figure 0
In Figure 0, you can see how a 10% increase in downloads plays out in a show that averages about 3000 initial downloads per episode and a show that averages 75.
The show with 3000 average downloads would see 300 new downloads if its audience increased by 10%. I would feel pretty good saying that 300 extra downloads represents meaningful audience growth.
Now, the show with 75 average downloads would see 8 (rounding up) additional downloads with 10% growth. And while the percentage across both shows is the same, I wouldn’t say with confidence that 8 more people are listening to the show or that I could infer a reason for 8 new listeners downloading an episode. But there are many ways to account for 8 additional downloads (e.g., downloads to multiple devices, stopping and starting streams in different apps, etc.)
Obviously, any new person listening to any show is a good thing! But we need to be very cautious when it comes to assigning meaning to change that’s statistically insignificant. That’s hard when a show is very niche, new, or just not made for a larger audience. Yet, it’s critical to thinking critically about the numbers you see on your dashboard.
All that said, though, you came here to read about stats. So let’s do it.
In the next section, I’ll go over how to think about:
Monthly downloads
Episode downloads
The lifecycle of an episode
Importantly, what I call the apples-and-oranges bias
Monthly Downloads
“Monthly downloads” is likely one of the first numbers you’ll encounter once you’ve logged into your podcast host. It’s the total downloads across all episode—new and old—over that calendar month.
Figure 1
Your monthly downloads might look something like this (figure 1). It's a bar chart that illustrates how many times your podcast was downloaded in a given month.
But what is this chart really telling us?
Monthly Download Trends
Intuitively, you expect to see your monthly downloads to be fairly even or maybe inching slowly up and to the right. But what you most likely see is something with dips and spikes that seem random—and jarring.
One month, it seems, you’re flying high. And another month, your downloads are in the crapper. And then, out of nowhere, you’re back on top a month or two later.
If you're like most podcasters I know, the first thing you think when you see a big drop is, "What the hell did I do wrong that month?" And then, when you see the big increase, you think, "Wow! What did I do right that month?"
You assume that fewer downloads equal bad and more downloads equal good.
But the truth is much more complex and likely has nothing to do with whether you were doing good things or bad things. And the reason is what I'll call the apples-and-oranges bias.
The Apples-and-Oranges Bias
When we say, “Oh, that’s apples and oranges,” we mean that we’re trying to compare things that are fundamentally different—even if they share a few traits. Sure, apples and oranges are both fruit. But pretty much everything else about them is different!
When we look at podcast stats, we see units of measure and categories of metrics that seem to be comparable. For instance, the number downloads (unit) in a month (category) seems like it should be comparable to the number of downloads in a different month. And sometimes they are. But more often, they’re apples and oranges.
Because podcast stats obscure the differences inherent to different metrics, we need to be extra cautious that we understand what we’re comparing before we start drawing conclusions.
For example, July and August seem like very similar months, right? They're both summer months. They're both 31 days long.
And maybe when you look at your monthly downloads, you see that your show was downloaded 5,000 times in July but only 1,000 times in August (figure 2). What the heck?!
Taken at face value, that doesn’t look good. But imagine that your show released four new episodes in July but went on hiatus in August and didn’t release any new episodes.
You can't make a one-to-one comparison between the two months when it comes to the number of downloads you see because there is something fundamentally different about their makeup. New episodes make up a disproportionate amount of downloads in any month. So any month that has no new episodes will have substantially lower downloads.
Figure 2
“Each month’s number of downloads is made up of 3 different categories of downloads: new episodes, recent episodes, and archive episodes.”
Here’s another way to look at the apples-and-oranges bias in action.
Each month’s number of downloads is made up of 3 different categories of downloads: new episodes, recent episodes, and archive episodes. When new episodes are released, they will account for a disproportionately greater proportion of the month’s downloads than recent or archive episodes. The reason for this is that new episodes are typically downloaded to subscribers’ devices automatically. Most people who subscribe to your show will receive the new episode within about 48 hours of its release. However, recent and archive episodes must be actively downloaded or streamed to count as a download.
Recent episodes will likely be downloaded more times than episodes further back in the archive simply because of their proximity to the top of the feed. Older episodes will still get downloaded, but each individual archive episode will account for a fraction of the downloads of new or recent episodes.
So months with different release schedules or listening conditions are apples and oranges. There are things to learn by looking at those numbers—but you can’t compare them side-by-side without context (figure 3).
Figure 3
Similarly, if I look at a day-to-day breakdown of my podcast’s downloads, I see that, right now, Thursdays are always spiky because that's when I release new episodes. I may also see a spike on some Mondays because I release extras that day. I'll see dips Tuesday and Wednesday, but not as big as the dips I see on Saturday and Sunday. Every day of the week is a different fruit!
Yet, our brains are wired to see patterns even—and this is important—when they're not there. A dip, a spike, a steady incline, or a steady decline all seem like they mean something because they appear to either be patterns or related to patterns.
We think we're seeing something like apple, apple, apple. But really, if we look just a bit closer, we see apple, orange, kiwi.
“Our brains are wired to see patterns in data even when they’re not there.”
So, let's take a closer look.
Download Math
When I see a monthly downloads chart that looks like figure 1, I don’t assume that something horrible happened to listenership in August and September—two months in which it looks like downloads cratered after rising steadily.
Instead, I remember the apples-and-oranges bias and consider what may be fundamentally different about those months than the months before and after them.
Immediately, I assume the drop was caused by the podcaster taking a break. No new episodes mean dramatically lower downloads (figure 4). And that’s fine! In fact, the existence of any number of downloads that month tells us something interesting and encouraging.
Those downloads are driven by people proactively going in and downloading old episodes. These are real live, curious listeners driving those downloads!
On a website, older blog posts typically see a small amount of traffic from search engines or social media. That's meaningful on its own. But with a podcast, where practically no automatic discovery mechanisms exist, that back-episode action demonstrates interest and intent. People downloading old episodes went back into your archive to find something to listen to. That's pretty incredible if you ask me.
Figure 4
When you're releasing new episodes, downloads of those episodes will make up the majority of your monthly downloads number (assuming that you don't have a huge back catalog). When you're not releasing new episodes, you haven't lost those downloads—it's just that there was nothing except for the back catalog to download.
The next time you release a new episode, those downloads will still be there.
I break down two hypothetical months of downloads in Figure 4, albeit with grossly simplified math. The first month, with weekly new releases, shows how much new episodes account for total downloads. In the second month, on hiatus, there are no new episodes with outsized download numbers. Instead, there are four more episodes in the archive (from the previous month) and a corresponding number of downloads for the month overall.
The month with new releases is going to register as a spike. The month on hiatus will register as a dip.
So if you look at the Monthly Downloads chart again (figure 1), you can see why August takes such a dip compared to July.
Is this just wishful thinking? Nope. Let's say your podcast is losing subscribers. Well, that's much more likely to happen gradually. Getting hundreds of listeners to actively unsubscribe from your podcast en masse would require a real scandal. Instead, losing subscribers looks more like slow, gentle churn. And yes, it happens! But the chart would look very different.
Similarly, the jump up in downloads from September to October in Figure 1 is, most likely, not from releasing a real banger of an episode or landing a big-name guest. It's merely the product of releasing new episodes again.
All that to say: monthly download numbers rarely tell you anything you don't already know. They're the laggiest of lagging indicators. You already knew when you were on hiatus. And you already knew when you released new episodes. And yet, monthly download numbers cause lots of panic for podcasters.
Monthly downloads recap:
All months are not created equal!
To track your monthly trends, you need to confront the apples-and-oranges bias.
What are the different variables that could impact downloads in one month versus another?
Celebrate downloads of your archives during months when you’re not releasing new episodes!
Now, let’s look at the metrics for individual episodes.
Episode Downloads
Figure 5
Episode downloads account for each time an episode file was downloaded or streamed within a specified period of time.
I find considerably more value in looking at episode downloads rather than monthly downloads. That said, some tricks are still involved in understanding what you see there.
When you look at a chart of the number of downloads, you probably see something like Figure 5. There's the date an episode was released, the title of the episode (or the file name), and the number of downloads it's received.
Depending on your podcast host, the time frame for those downloads might be all-time, last 30 days, or first 30 days (or something similar).
All-time downloads are exactly what it sounds like—the number of times an episode was downloaded for the entire period it's been live. With all-time downloads, you typically see the number go up the further back into the archive you go. Episode 10 is likely to have more all-time downloads than Episode 25 simply because of the difference in how long the episodes have had to accumulate downloads (figure 6).
If you see the opposite trend, that is, new episodes have more downloads than older episodes, that's great! That is likely significant audience growth. If you see a big spike in downloads for an episode in the back catalog, that's something to take note of so you can think about what might have caused that interest.
Figure 6
But noticing that earlier episodes have more downloads than recent episodes does not mean your show is losing listeners. Let me repeat that. Older episodes almost always have more downloads than recent episodes simply because they've had longer to rack them up. It doesn't mean your show is losing listeners.
“Noticing that earlier episodes have more downloads than recent episodes does not mean your show is losing listeners.”
"Last 30 days" is a rolling time frame that is less than helpful. Typically, what you'll see when you look at the "last 30 days" time frame is strong numbers for your most recent 3-5 episodes because they were released during the time frame in question. Then, you'll see a significant drop-off in downloads because every other episode in the catalog is past that initial boost from subscriber downloads. This is perfectly normal and doesn't mean a thing.
"First 30 days," on the other hand, is an interesting time frame because it beats the apples-and-oranges bias in the other two time frames.
With all-time downloads, the apples-and-oranges bias causes you to forget that newer episodes have had less time to accumulate downloads. An episode released a year ago fundamentally differs from one released a month ago—they're apples and oranges.
With the "last 30 days" time frame, the apples-and-oranges bias causes you to forget that newer episodes have the benefit of the initial boost of subscriber downloads received in the first 24 hours or so. Older episode stats will look minuscule in comparison. An episode that's been released within the last 30 days fundamentally differs from one released outside that time frame—they're apples and oranges.
The "first 30 days" window, if your podcast host provides it, gives you a way to compare apples to apples (figure 7). Each episode has the same 30-day window in which to accumulate downloads. An episode released on September 1 will have stats that run to September 30. An episode released on November 1 will have stats that run to November 30. Once that full 30-day window has passed for two or more episodes, I can compare them—apples to apples.
Figure 7
In Figure 7, we’re comparing the “first 30 days” of Episode 10 with the “first 30 days” of Episode 25. Episode 10, released 18 months before Episode 25, received 250 downloads in its first 30 days live. Episode 25 received 350 downloads in the same amount of time. That means that over 18 months, the audience may have grown by more than 30%!
Episode Downloads Recap:
Older episode will (almost) always have more downloads to date than newer episodes
Time frames matter—looking at the first 30, 60, or 90 day-time frames gives you the best chance of comparing apples-to-apples
Next, let’s take a closer look at how an episode’s metrics will change over time.
The Statistical Lifecycle of an Episode
On the day an episode is released, most of your subscribers will have that episode automatically downloaded to their devices. Whether or not they listen to it that day doesn’t matter. You’ll see the download in your stats.
That means that the first 24-48 hours of an episode’s “life” accounts for a huge proportion of its all-time downloads (figure 8).
Figure 8
Over the next 4-8 weeks, that episode will still see more downloads than much older episodes. But barring some unexpected press, no single day or week will beat that first week of release in terms of total downloads.
I'll note here that Transistor.FM, the podcast host I use and recommend, recently rolled out the neatest little chart. In fact, it does something similar to what I tried to demonstrate in the chart above. It uses a rolling 30-day window to show how many downloads each episode received in its first 30 days, day by day. Using that chart, you can actually see that spike in the first 48 hours and then the long-tail decline in downloads over the course of the 30 days.
Figure 9
Statistical Lifecycle of an Episode Recap:
Episodes receive the biggest proportion of their downloads in the first 24-48 hours
Downloads of an episode tend to decline, day by day, over time
So what do you do with this information?
Honestly? Not much. Podcast stats require a lot of work to make sense of very little information. And very, very few podcasts get enough downloads to think strategically about the content they create. That sucks—but it's true.
Actually, I don't believe it sucks. I believe that it's a blessing to be freed from the false clarity of data.
Anyhow, the stats that really matter are the ones that you decide are important based on the strategy behind your show. Did you start your show to connect with cool people who like to talk about the same things you do? Did you start it to attract new clients or customers? Did you want to carve out some unique positioning or test-drive an idea for a book? Did you do it for advocacy or organizing reasons?
Based on why you started your podcast and what you hope to receive from it, there is probably some quantitative metric you can find that will be more helpful than whatever your stats dashboard says.
But I want to put in a good word here for qualitative data and metrics.
Tracking the numbers is not the only way to get useful information about your show and how it's working. Qualitative data are things like messages from listeners telling you what they thought of your latest episode or client intake forms that say that they found you when a friend recommended your podcast. They're things like good conversations and new network connections.
Culturally, we've developed a bias toward quantitative data and away from qualitative data. But that bias—like the apples-and-oranges bias—often leads to trouble. It gives us a false sense of clarity, a feeling that we know what's going on even when we know no such thing.
Qualitative data is crucial to understanding the value of your podcast.
Final Thoughts
Remember, podcast stats are weird.
If you're taking them at face value and going with the first flash of "understanding" you get, you're missing most of the puzzle. And you'll likely think your show is doing "worse" than it is.
When it comes to podcast stats, or any data and metrics for that matter, look for the story within the story. It's always there waiting for you. And it's a good one.