How to find your TRUE top podcast episodes

I think I've cracked the podcast data code. Here's how.

Jan 13, 2025

I think it’s time to rebrand January.

Most people talk about the January blues, the post-Christmas slump, the month where nothing happens.

What they don’t realise is that podcasters get a second Christmas because THE YEAR’S PODCAST METRICS ARE COMPLETE!

When the calendar ticks over into a new year you now have a full 365 days of podcast data to play with from the previous year. It’s just like Christmas all over again!

I love to review a whole year of data, and in particular looking at which episodes have performed best (and worst), because it helps me understand what’s worked and what hasn’t, what we should do more of and what we should stop.

I spent a happy day on my first week back at work diving deep into our previous year’s podcast data, and I’ve started experimenting with a few new analysis techniques which have garnered some really interesting insights.

When I hit on what I’m about to share with you, I genuinely felt like I’d cracked the podcast data code.

So, below I’ll share my all-new step-by-step process for reviewing a whole year of podcast data to find our true top episodes.

Spoiler: it’s not (just) about downloads.

I’d also love to learn from you: it would be great to hear how you analyse your data and how you hone your content strategies. Let me know by reply to this email or in the comments on Substack. Talk nerdy to me.

Know someone who loves diving into podcast data? Share this email with them (it’s free)

Beyond the download

2024 seemed to be the year that podcasters started to seriously move away from downloads as the primary metric of success.

Perhaps this was driven in part by Apple’s move to change the way Apple Podcasts treats downloads in the iOS 18 update.

Whatever the cause, the shift is a good one. While downloads remain useful to a point, they are at best a blunt instrument: a download of an episode (which your hosting provider has to interpret from a number of different actions - automatic or manual - within multiple different podcast listening environments) does not necessarily equal a listen.

They also tell you nothing about how long users are listening for, or whether they come back for more.

I’ve used downloads as a key metric in previous years. But in 2025 I want to get more specific. I want to know how many people are actually listening, how long they are listening for, and how loyal they are.

I work within an organisation which is refocusing its efforts on driving paying subscribers. So loyalty and engagement metrics are going to be increasingly important to help us understand our listeners’ propensity to subscribe - and to target our subscription marketing messages accordingly.

Harvesting the data

We host our podcasts with Acast. They are fabulous in lots of ways but, like most RSS hosting platforms their metrics are limited to raw downloads.

To drill down into actual user behaviour we have to get data from the listening platforms themselves. Apple Podcasts and Spotify both provide useful analytics themselves, through Apple Podcasts Connect and Spotify for Creators respectively.

Fortunately these two platforms account for the vast majority of podcast listening (65-70%, in our case). For the purposes of this analysis I’m happy to take these as our data source and extrapolate.

Getting the data itself is pretty straightforward: simply log in to your account in the relevant platform, navigate to the analytics page, set your time window to the previous year and export as a .csv file.

This can then be imported into Excel or Google Sheets for processing.

Frustratingly, Apple and Spotify don’t provide the same metrics. There is some overlap, but they also use different words to describe similar things.

They do, however, provide a little help: if you hover over the relevant metric in the platform, a little explanation will pop up. I’ll also look at doing a deep-dive on the different terminology in a future issue of Podcast Strategy Weekly.

Making it useful

Seeing the data from your various sources in one place is immediately useful. I enjoy filtering or sorting by different categories to see which episodes are doing well on different metrics. This can teach us some useful lessons about how to optimise for specific things.

Say a particular episode over-performs on listen time. We can go back and listen to that episode to understand why, and try to identify best practices that we can roll out across future episodes to improve our watch time performance.

So far, so obvious.

But looking at a sheet with loads of different metrics can get a bit overwhelming. We need to simplify - particularly if we’re planning to communicate around a wider team who maybe aren’t as geekily fascinated with all the minute details.

Ideally, I want to be able to quickly and easily see which are my overall top performing episodes from the year just gone, so I can emulate the best of them in the year to come.

Previously I’d do this by simply looking at which episodes had the most downloads. If we’re trying to move beyond this, things get more complicated.

Now we’ve got loads of different metrics in our spreadsheet. Sorting by individual columns just gives us the episodes that performed best for that metric - that’s not the same as an overall ranking.

How do we work out our top performers?

I’m not a data scientist. If you are, maybe the answer is embarrassingly obvious. But it took me a lot of head-scratching (and some stupid questions to chatGPT) to figure out a solution.

When I finally hit on the following technique I felt like I’d struck gold.

So for those of us who aren’t data scientists, here’s how to do it:

Step 1: Normalise the data

First, we need to “normalise” the data in each column.

The columns all use different scales so we need to standardise them in order to make a fair comparison.

To do this, add a column in between each metric. You’re going to use the following formula to standardise your data:

Normalised value = (Actual Value - Min Value) / (Max Value - Min Value)

In spreadsheet-speak that looks something like this (using cell B2 as an example, in a column of 100 entries):

=(B2-MIN(B$2:B$100))/(MAX(B$2:B$100)-MIN(B$2:B$100))

(The $ signs lock the reference cell, so you can copy and paste without messing up the formula)

Copy the formula across the whole column, then repeat in the rest of the blank columns you just added.

Step 2: Assign weights

Now you need to decide how important each metric is to you. This will vary depending on your strategic goals. In my sheet, I decided to rank episodes for total listeners (Apple + Spotify), Average listen time, average listen through (percentage), and downloads.

I ranked these in priority order, giving a number to indicate how important they are to me. The list looks like this:

Total listeners: 10

Average listen time: 8

Average listen through: 5

Downloads: 3

For me, the total listeners is the most important metric because these are people who have actually heard the podcast, not just downloaded it.

I’ve prioritised listen time next because that indicates how long people have been willing to spend with our brand.

Listen through gives an indication of how much of an episode people have listened to, but it’s relative to the episode duration. It’s useful to indicate how engaging the episode is, but for me not as important as total listen time.

Downloads is still on the list because it’s a signifier of overall podcast health, but its value is limited, as I described above. So it receives the smallest weighting value.

Step 3: Apply weighting

Create a new column and add up the normalised metrics, multiplying each of them by the weighting to give a weighted score. The calculation looks like this:

Weighted score = (normalised metric 1 x weight 1) + (normalised metric 2 x weight 2)… and so on

In your spreadsheet that might look like:

=(C2*10)+(D2*8)+(E2*5)+(F2*3)

Copy this down the whole column.

Step 4: Find your top episodes

Now simply sort your sheet by the weighted score column in decreasing order (Z-A).

Et voila! You’ve just revealed your top episodes by a combination of the metrics that are most important to you.

The results

Doing this process for the podcasts I manage revealed a couple of interesting outliers.

This new ranking system pushed one particular episode to the #3 slot - but if I’d been sorting by downloads alone I might have missed it entirely. The episode - an extended interview - had performed averagely in terms of downloads, but had an absolutely massive average listen time of over 60 minutes.

Similarly, it turned out our top performer by downloads didn’t even make the top 10 by weighted score. It had a huge number of downloads, but less than 20% converted to actual listeners. This suggests the download number was an anomaly, perhaps resulting from auto-downloads or similar.

Understanding which of our episodes performed best over the past year can help us target our efforts more strategically in the coming year. It helps us learn what to do more of, and what to cut. Using the techniques above will help you do to this in a more nuanced and accurate way than simply relying on download data.

Have fun with it! And don’t forget to let me know if you have any better techniques.

Enjoyed this? Why not share it with a friend, colleague or fellow podcaster?

Download your year’s data from Spotify, Apple and your hosting platform
Import into a single spreadsheet and rank using the system explained above
Look at your top performing episodes from the past year based on your weighted rankings. Which best practices can you apply from these episodes into your content in the coming year?

Worth your time

Matt Deegan’s newsletter, Matt on Audio, is an exceptionally high-value read. In the first days of January he released a series of emails discussing the relationship between podcasts and video. It’s a fantastic piece of work which eloquently addresses the surprisingly complex and nuanced discussion around the value of video for podcasting.

In particular, I liked this quote from the second email in Matt’s series:

If you are in the long-form conversation business, I think it’s essential to be focused on your potential consumers and how they spend their time. To be successful, you need to distribute your content in the ways that best appeal to them and make them more likely to consume it. If you’re trying to sell a product, you want it to be in every store and not just in Sainsbury’s (no matter how successful it is doing there).
If someone does not listen to podcasts but watches YouTube videos in your genre, should you be trying to teach them to try podcasts, or should you put your content in a place they are already visiting?

Read the whole of that email here (and check out the rest of the series while you’re at it):