Understanding Your Download Stats
Introduction to Stats
Podcast stats represent a window into your listener behaviour. Are your downloads growing over time? Is a certain episode particularly popular? Are your listeners are clustered in a single geographic region of the world? Your podcast stats give you insight into this data, based on how your listeners computers and smartphones identify themselves when they download your podcast files.
Stats Collection 101
When you publish a podcast, your podcast has an RSS feed file that includes a list of your episodes, and the address of a server on the internet that hosts each of your episode audio files. When a listener chooses to download or listen to an episode of your podcast, their podcast app connects to the hosting server and requests that episode file. During this request, your host can gather some limited information about the listener, and the collection of all of these requests makes up your podcast's stats.
When an episode file is requested, the first piece of information is the request itself – you know that one additional computer somewhere has requested that file for download at the time the request came in.
When a computer makes a request for a file over the internet, it also typically sends along some additional information (called the "user agent"), which notes software used to make this request. For example, a request made by iTunes will identify itself as coming from iTunes, and a request coming from Chrome will identify that it is coming from Chrome (along with some additional information, such as the computer's operating system). While this user agent info is a common norm, it is not enforced, and requesting software can identify itself however it wishes (often called "spoofing" if the user agent is inaccurate), or it can simply leave the user agent field blank. One common form of spoofing is podcast apps that identify themselves as iTunes (despite not being iTunes), in order to ensure that the hosting server serves out a file as it would if a user was requesting the file from iTunes.
Additionally, when making a request for a file, the requesting computer provides an IP address to send the file back to. The IP address represents a specific computer on the internet that would like to receive the file.
The most direct representation of your listener count is the number of download requests a podcast episode has received. That said, it's not a perfect 1:1 measure, for a couple of reasons.
First, a request simply identifies that a computer has requested the file, and not that a listener has listened to it. Many podcast apps allow users to subscribe to podcasts and automatically download episodes as they're published, regardless of whether users listen.
Second, the same computer may request the same file multiple times. This can occur for many reasons, but one of the most common reasons is if a user elects to stream your podcast, which means their computer or phone listens to it as it downloads from the server (rather than downloading the entire file in advance). To save data, many apps only request a small portion of a file while streaming, and then make a new request for another portion. As a result, a single listener may show as more than one download.
Removing these multiple streaming requests and tallying them as a single user is also not particularly reliable, as many requests from a single IP address could represent multiple users who are on a shared network – whether at a school, or on a corporate network, or connecting through a single VPN endpoint.
Because a download request includes the requesting computer's IP address, it's possible to roughly approximate the geographic location of the computer making this request. This is an inexact art – as IP addresses shift; a block of IP addresses may all be assigned to a cell phone provider in a single city (and thus show up as coming from that city), however the users assigned those IPs may be far from their provider's head office.
Further, users may elect to connect to the internet through a VPN (whether for corporate, privacy or speed reasons), which can mean their requesting IP address could be from whatever country their VPN endpoint is located in, rather than their own location.
While we caution against treating location data as 100% reliable, we believe that country-level demographics (as we report in Cast), strike a good balance between reliability and specificity.
Similarly, it's difficult to infer much in the way of demographic information about your listeners. In the case where their IP address is correct, and their user agent is accurate, you will have accurate information about their country of origin and the device they're using to download your podcast. Beyond that, however, it's difficult to reliably infer more information about these listeners.
While podcast stats are in many ways limited, they can still provide valuable insight into your listener behaviour, in particular when comparing data over time (are my listeners increasing). For further demographic information, you may find it's valuable to put together a short survey and ask your listeners directly.