You may notice that subscriber analytics can only be viewed with a granularity of days. This is intentional, and has to do with how subscribers are measured.

Background

Subscribers are measured by counting the number of unique downloads of a feed in a day. If a subscriber's podcast app checks your feed every hour, they will still only be counted a single time.

Let's say this subscriber shuts their phone off. Their podcast app will stop refreshing your feed, and there will be nothing to count.

Why smaller bucket (e.g., hourly) analytics are unavailable

There are a few reasons:

  • When your subscribers' devices are off or have no internet connection, they will not appear in, say, an hourly bucket. Hour by hour, subscribership numbers would fluctuate by how many subscribers phones or computers are actually able to connect to the internet. In measuring this, we found it to be very significant.
  • Some podcast apps only refresh feeds one or two times each day. These subscribers would not appear in most of the hourly buckets, making their subscribership invisible.
  • The usefulness of better-than-daily subscribership numbers is dubious. In surveying podcasters, we did not find any demand for the feature.

Why larger buckets (e.g., weekly) are unavailable

There is one main reason for this. The idea of a "unique download" of your feed is based on a fingerprint of the app downloading the feed. This includes information like IP address and User Agent string. Within one day, it's unlikely that these will change. Across the span of a week or month, it's very likely that these things will remain constant (IP addresses change, apps get updated with new User Agent strings, etc.) meaning some subscribers will be counted multiple times.

Pinecast could create what's known as an aggregation, which means that we would perform an operation like averaging each of the days in a larger bucket. One week's subscribers would be, say, the average of each of the days' subscriber numbers within the week.

Aggregations are problematic because they masks trends. Averaging each day in a week or month would hide growth in subscribers by pulling down recent "high" subscriber numbers. For instance, taking the maximum subscriber count in a week (rather than averaging) would hide downward trends in subscribership. Taking the minimum count in a week would bias towards the weekends (where subscribership is usually lower), hiding growth on weekdays.

Rather than impose these tradeoffs, we show only daily subscriber counts. If you have an idea for how to avoid these problems, please don't hesitate to get in touch.

Did this answer your question?