This article originally appeared in issue #2 of Podertainment Magazine
Author: Todd Cochrane CEO RawVoice Parent company of Blubrry.com
Hey folks! We’re going to get into the nitty-gritty of podcast statistics data. We know how confusing some stats can be, this is where I break it down for you.
I want to share a little history with you before I get into the topic at hand. I started my own podcast in October 2004 and as near as I have been able to figure it, my show was one of the first 50 or so podcasts to launch. In the early days, we were not worried about stats — our biggest challenge was bandwidth. There were no so-called unlimited bandwidth services.
Those hurdles were quickly overcome, and within a year we started focusing on who was listening, where and how. Today we have vibrant data that can tell us exactly how many people listened on the Web, used an app, or kicked back and watched on a set-top-box. Tracking the all of that now is an afterthought.
Two questions remain: Did they really listen and For how long? I will share with you insights no one has to date revealed, data that is now 100 percent capable of being measured and will benefit you as you grow your show and your audience.
In a related article I provided some statistics that raised a lot of questions. We figured that might happen. This time, I have refined and broken down the data into chunks that are comprehensive.
With the RawVoice/ Blubrry podcast statistics reporting that we do for our clients, on every download / stream request, or “hit,” we analyze in great detail whether the hit should be counted as a download. With this precision analysis we can see out of range trends and account for that, as percentages fluctuate daily, show to show, episode to episode. With our proprietary algorithms we provide consistent podcast download totals regardless of what these percentages are on any given day.
Given a random snapshot in November 2013, of 5,606,161 hits for media files ranging in size, we determined:
Looking at the 55.9 percent of hits that were not countable, 22.4 percent ip duplicate requests, mostly caused by iTunes (iTunes may send one request to get the file size, then a separate to download the file). The remaining 33.6 percent stem from a variety of issues, such as invalid HTTP status code, empty byte range request size, non-existing files, invalid file name (non-podcast media), etc.
When we see partial download requests, we take the time to assemble the bytes requested to determine how much of the file was downloaded. In some cases, we will only see one or two requests from the same unique visitor, and in other cases we see hundreds of small byte range requests from a unique visitor. Factoring in the byte-request data we know exactly what portion of each file is requested, allowing us to calculate how much of the file was downloaded. Of the 19.2 percent partial file downloads they make up small segments aka repeats of portion of a file already counted we do not count those partials as they have been accounted for in the other confirmed downloads.
We reported 24.2 percent as counted downloads (underlined totals), with 22.5 percent being from unique IP addresses, leaving only 1.7 percent of the hits accounting for two or more downloads. The end result shows us that two or more downloads coming from the same IP address is very low.
Any given day, the percentage breakdown of countable and not-countable downloads fluctuates based on many factors. If you rely solely on download hits, you will find that your numbers will not be consistent and will not reflect your true download total or audience size.
It is important to note that unique IP address data is critical validation data for podcast media downloads. Based on the results above, we know that the final download / stream total will always be higher than and relatively close to the unique IP address total. We (Blubrry/RawVoice) knew this back in 2005 when we started measuring podcast downloads, and as you see from our small snapshot from November, it still holds true today.
If you are concerned about being audited or need accountable details to meet the Sarbanes-Oxley Act, make sure your podcast measurement is taking the necessary steps to calculate the true download total.
Here is where I am going to get up on my soap box for just a moment. If you are counting your downloads with anything other than a trusted podcast statistics platform, you are likely over-reporting your total audience size. It’s complicated stuff. Get the unique IP count correct by throwing out the garbage and you will find that Unique IP is a large indicator of true audience size. This is evident, as we’ve noted above, because we do not see Unique IPs coming back very often for the same file.
In a future article, I will translate the percentages into describing what we are seeing in an actual listening rate. With Byte Range being used nearly exclusively in the space, we can tell you based on server data just how long they are listening to your show.
If you desire your podcast network to get on track and have media statistics that keep you honest, feel free to reach out to me at firstname.lastname@example.org. We host and measure the biggest podcast networks in the space