One big number? How to combine audio + video podcast data across Apple, Spotify, and YouTube
Here's a question I get regularly:
"Our podcast audience is fragmented. Some people watch on YouTube. Others play on Spotify. Still others listen in Apple Podcasts. How am I supposed to add that all together? My boss doesn't care about the differences between views, starts, streams, and plays. I just need one big number to share with the team."
As podcast consumption patterns evolve and the download's relevance continues to decline in an era of YouTube consumption and Spotify pushing video podcasts (neither of which generate meaningful download traffic), combining podcast consumption data across platforms has never been trickier.
Here at Bumper, we understand the need for a grand total. One big number for leadership, or sponsors, or on-mic talent.
And indeed, it's possible to arrive at one big number. But the details matter.
Today, I want to share Bumper's approach to adding together numbers that were never really meant to be combined: video + audio consumption across the three big consumption platforms. But first, an important distinction…
Delivery vs. consumption
Downloads are a measure of podcast delivery. They measure how many times audio or video files were requested by listeners' devices and delivered by a hosting provider. But just because a file was requested and delivered does not mean it was actually played back.
We often see a sizeable gap between the number of files that get delivered (measured in downloads) and the number of files that actually get played. Apple's 2023 changes to automatic downloads narrowed that gap, but for many shows, we still see a significant delta between delivery and consumption. Downloads aren't people, and not all downloads get played.
It can be very tempting to combine consumption data and delivery data. For example, I know many people who want to add together YouTube views + podcast downloads. Or to add up Spotify streams + podcast downloads. I typically recommend against this approach. Why? Because YouTube views and Spotify streams are consumption metrics, and downloads are a delivery metric. Combining delivery and consumption metrics can be misleading, because delivery is not consumption.
Bumper tends to focus on consumption metrics, like verified listeners, listen time, and actual podcast playback. Consumption metrics are a far better way to gauge true audience size and engagement, and they tend to be less volatile than delivery metrics. Remember the changes Apple Podcasts made to automatic downloads with iOS 17? A great way to guard against future changes to automatic downloads is to shift your team's focus to consumption metrics.
And there's good news: there are some very reasonable ways to add together consumption metrics across platforms, across both video and audio, to get one big number.
But again, the details matter. A lot.
Let's start adding some numbers, shall we?
YouTube views
Many people feel familiar with YouTube views and what they represent. But what exactly is a YouTube view? YouTube's documentation offers a surprisingly vague explanation:
A view occurs when a person watches your video. In order to preserve accuracy in view counts, irregular playbacks (like spam) are removed from the public view count metrics. In incrementing view count, YouTube algorithmically determines user intent.
YouTube does not disclose exactly how their system calculates view counts, but industry estimates suggest:
A YouTube user must watch a video for a minimum of 30 seconds to increment a video's view count
A single user can generate multiple views per day, with a maximum of 4-5 views per day
This makes intuitive sense. If a human being watches a YouTube video for at least some minimum threshold of time, it counts as a view. The same person can watch the same video more than once, and there's a reasonable cap on the number of views someone can generate in a window of time.
In other words, YouTube views are a consumption metric that corresponds to "unique human-triggered video play initiations that meet a minimum playback threshold."
Fair enough. So then, what's the audio podcast equivalent of a YouTube view?
Spotify streams
Very helpfully, Spotify offers a metric that is quite similar to YouTube views. Spotify calls them streams:
We count a stream every time a listener or viewer streams your show for 60 seconds or more on Spotify
Streams are similar to YouTube views in that they have a minimum listening threshold. They're also similar to YouTube views in that a single Spotify user can generate multiple streams for the same episode (for example, when someone revisits the same episode in a new playback session, or after they play another piece of media on Spotify).
For this reason, even though they're not identical, I feel comfortable adding YouTube views and Spotify streams together. They're both consumption metrics that correspond to "audio or video playback, initiated with human intent, that meets a minimum playback threshold."
So far so good, right? Well, get ready for a curveball...
Apple Podcasts plays
Apple Podcasts Connect reports a metric they call plays, and it's perhaps the most commonly misunderstood podcast metric I've encountered.
Given a plain-English interpretation of the word, Apple Podcasts plays seem like they should be roughly equivalent to YouTube views or Spotify streams. But they're not. Apple's definition of plays:
The total number of times people pressed play on your episode
Read that one more time... closely:
The total number of times people pressed play on your episode
Apple Podcasts plays reflect the total number of times an Apple Podcasts user clicks or taps the play button, including when they resume playback of a paused episode. If you play an episode… then pause it… then resume it… then pause it… then resume it… you will have contributed to that episode’s play count multiple times.
A single user, during a single playback session, can generate many play actions. This is why Apple Podcasts average play counts tend to correlate closely with episode duration: longer episodes tend to get paused and resumed more times than shorter episodes.
For this reason, I don't spend much time on Apple Podcasts play counts. And I discourage teams from including Apple Podcasts plays in a tally alongside YouTube views and Spotify streams because they're so different. As the song goes, "one of these things is not like the other."
But this presents a problem: if Apple Podcasts doesn't offer "unique human-triggered play initiations that meet a certain minimum playback threshold" then how are podcasters supposed to include Apple Podcasts consumption data in our tallies?
Here's what Bumper does: we estimate.
Our estimates include a few ingredients, all sourced from Apple Podcasts Connect:
Episode-level Apple Podcasts all-time total listeners
Episode-level Apple Podcasts listeners by day/week/month
Episode-level Apple Podcasts all-time episode retention (AKA dropoff curves)
We use the retention data to estimate the percentage of listeners who spent at least 60 seconds with an episode, then we compare the sum of daily/weekly/monthly episode listeners to the episode's all-time listener number, then we scale daily/weekly/monthly by those two scaling factors. This results in both an all-time "estimated plays" metric and a daily/weekly/monthly "estimated plays" timeseries.
Is it perfect? No. It's an estimate. But it's the closest we have to "audio or video playback, initiated with human intent, that meets a minimum playback threshold" for consumption that happens within Apple Podcasts. And it's the number we feel most comfortable summing together with YouTube views and Spotify streams.
At Bumper, we call this combination of views, streams, and estimated plays as simply "plays." We use this term because it lines up with the plain-English meaning of the word, and because industry experts suggest using “play” to mean times “when we know a human being has pressed the play button.”
And starting this week, the Bumper Dashboard will begin to display it for episodes and shows, both as an all-time number, and by day/week/month.
I would love for Apple Podcasts Connect to report a play count that more closely aligns with YouTube views and Spotify streams. Were Apple to do this, three of the largest podcast consumption platforms would offer similar-enough metrics that our industry could have a real conversation about moving towards consumption-based metrics (vs. delivery-based metrics) as the currency we use to transact.
What about other podcast apps?
So far, I've only mentioned three of the big podcast consumption platforms: Spotify, YouTube, and Apple Podcasts. While these services do not represent the totality of podcast consumption, they make up the lion's share of consumption among most of our clients' shows.
There are many more podcast apps out there, but few beyond the big three offer publishers consumption-based playback measurement. If you work on a podcast app that offers creators consumption-based play counts similar to YouTube views or Spotify streams, I would love to hear from you.
Not all plays are equally valuable
So, it's possible to add views, streams, and estimated plays together. But that doesn't mean they're equally valuable. Quarters, dimes, and nickels are all coins. But they have different values.
For example, among many podcasts we work with, YouTube average view duration is lower than Apple Podcasts average consumption. Even though YouTube view counts might be much higher than Spotify streams, it's not uncommon to see more total time spent on Spotify than YouTube.
So if part of your podcast goal is for people to spend time with your show, I recommend looking at play counts alongside listen time. Because if you're optimizing for time spent with your show, the reach/engagement tradeoff looks very different across platforms.
Plays in the Bumper Dashboard
The Bumper Dashboard is Bumper's measurement aggregation software. It's available to all Bumper's consulting clients, and we recently starting offering invitation-only access to non-consulting clients. The same type of play counts I've described here are available at both the show level and the episode level within the Bumper Dashboard.
My colleague Jonas and I will be at Podcast Movement Evolutions from March 31 to April 3, 2025. We'd love to meet up and compare notes about podcast measurement, and hear how your team manages to combine the various metric across platforms, and across video + audio. If you plan to be in Chicago for the conference, please do reach out.
Remember
Podcast delivery metrics are different from podcast consumption metrics. Avoid combining them to avoid confusion.
It is possible to combine video and audio metrics together across platforms, but great care should be taken to consider platform-specific idiosyncrasies. Add numbers responsibly.
YouTube and Spotify both offer similar consumption metrics that correspond to play initiations that meet a certain playback threshold. The plays metric in Apple Podcasts Connect can be easy to misinterpret, as it does not align with YouTube views or Spotify streams.
It's possible for podcasters to use Apple-supplied episode retention data and listener counts to estimate an Apple Podcasts equivalent to Spotify streams. This is how Bumper displays total plays in the Bumper Dashboard.
As downloads continue to become less relevant as a reach measure, the podcast industry should shift its focus to consumption-based metrics.
One big number is possible.