I used to check our numbers every day, at 12:30. It was my ritual to eat lunch, and then take care of a little bit of network stuff before returning back to work. I would check the daily download totals, and dutifully mark them down on a per show basis. For the year of 2012, I can tell you exactly how much every show was downloaded on any given day.
But putting that much time and attention toward keeping track of the numbers with that level of granularity became untenable. As we added more shows, I started checking them less and less. The problem is, you really don't need that much data in a spreadsheet... especially if your podcast download service is keeping track of things for you.
So... How much is too much? How little is too little? Which metrics should you pay attention to, and how much can you synthesize from those numbers? This won't be the definitive guide, but it should give you some idea of what my approach is to tracking our numbers.
Where the numbers come from
In a previous post, I outlined which podcast hosting services we've used over the years. The majority of those, like SoundCloud and Libsyn, will give you download metrics as a part of the service. For others, you will need to use a service like Podtrac to get numbers. Each of these counts downloads in a slightly different way, but what's important is consistency. Most of them also let you pull a specific report for a certain time period.
Day One Episode Downloads
When an episode is released, I want to know how many people download it in the first calendar day. This actually does a pretty good job of showing how many people are subscribed to the show... When the post goes live, people's podcatchers will immediately grab it.
Something worth noting here is that subscriber counts are useless. Services like Feedburner and FeedPress will spit out a subscriber count, but that number is skewed and bloated by the number of RSS-crawling robot spiders that exist. Numbers are only useful if they represent real actions taken by real people.
Weekly Episode downloads
One key number I look for is how many times an episode was downloaded in its first full week of release. This number works great to show you short-term growth (or shrinkage), pinned to decisions that were made in the previous show. What game did we cover? What was the length of the episode?
Determining this number could be really simple, or slightly complicated. The simpest version of this is just selecting a 7-day period after release and recording the number. The complicated version (which I do) involves pulling numbers from an Early Release feed (which goes out to Patreon backers), the number of plays on SoundCloud (a number that incorporates the people who download episodes via RSS), and the number of direct downloads.
These numbers are added together to get the golden number. I then do some spreadsheet math to determine episode-over-episode growth or shrinkage on a per-episode basis. But that's only a small part of the picture.
Monthly Episode Downloads
Not everyone grabs an episode right away. Downloads stay strong for about 30 days after an episode is release (explained further: It takes about 30 days for an episode's daily download total to stop exceeding a 1% marginal change over the current total of downloads). This means the 30 day total is important, too.
Getting this number is simple. You just do the calculation for the weekly download total, except for the 30 days after release (screw you, 31 day months!). I then run a percentage change comparison against previous episodes. Usually, the deviation is smaller at the monthly scale than the weekly scale. I don't know if this is regression to the mean, or what.
Episode Listener Totals
Even a monthly download total won't get you a full picture. Those fluctuations can't be used as a solid indication of how many people are actually listening to a show. Sometimes people miss an episode, or skip it for some reason or another.
To put a kind of "softening" filter over the numbers, I take a snapshot of how many listeners a given show has by taking an average of the 30-day downloads for the previous quarter's worth of episodes (12 for a weekly show, 6 for a biweekly show). Even though this number takes more processing to arrive at, I think it's more reliable. Freed from the daily, and even weekly fluctuations in a show's downloads, you get a number that grows or shrinks steadily. It's a more realistic and sound figure to make decisions by.
Monthly Show Listeners
Once all of the listener totals are in for a given month (this will happen at the end of the next month), I take an average of the Listener total for each episode released in that month, and then mark that down as the number of listeners the show had that month.
Monthly Show Downloads
Now we're back to simple numbers.
At first, I kept track of downloads every single day. Then I moved to calendar weeks (Sunday to Saturday). But even that proved to not provide any real insight. Now, it's my ritual to grab a total of a show's downloads at the end of every month. This is just a repeat of the process for getting weekly downloads, except with more days.
Downloads of new content, or "fresh" downloads, only account for a small portion of the total downloads for a show in a month. It's great to know how many "backlog" downloads you have, to show you how many people are going back and listening to episodes older than a month. If this number-- as a percentage of total downloads for a month-- spikes, then you know that new listeners are doing deep dives on content thats new to them.
The perfect way of calculating this is really complicated, I use the easy way. I take the 30 day downloads for a given month, then add those up. Those are the "fresh" downloads. Then I subtract that number from the total downloads for that month to determine the "backlog" downloads. These are then expressed as a percentage.
For reference, evergreen shows have a higher percentage of backlog downloads. WOFF and BSC usually run at around 75% backlog downloads (i.e. 75% of the downloads go toward episodes that are older than 30 days), while something more periodical like The Level stays around 30% backlog downloads.
Take the monthly downloads for the three months that make up a quarter, then add them. That's the quarterly download. There really isn't much you can do with this figure, but it's nice to have.
What's it all for?
The release of our first Bloodborne episode of Bonfireside Chat inspired me to write this post. It's been a huge success for us, and within seven days it became our most-downloaded episode of any show, throughout the history of the network.
I was also inspired by the thought process that led me to write my previous post, on energy. These numbers go up and down. In general, they trend upward, but if I got bummed out every time they dipped I'd be in serious trouble. "Non-attachment to results" means doing the best work you can, but not defining yourself by how it's received.
Most publications keep numbers like this in order to determine advertising rates. We only do small ads for listeners, but even those are determined by a CPM calculation against the previous quarter's downloads for a show. There's a very loose correlation between downloads and Patreon dollars, but that could also be a function of time.
These numbers don't directly make us money, and they don't really jazz us up between episodes. So why do I keep track of them? Because they might be useful. They give a quick view of the health of the network, and confirm that we're slowly and steadily getting more listeners and more attention. There's no big trick to "growth hacking" this kind of stuff in the long term. We've said it time and again, the real magic is showing up. As long as we keep showing up, those numbers will go up.