Impolitic Life

Privacy, accountability and podcasting

<![CDATA[Not Too Geeky » Audible’s service – trouble ahead for podcasters?: Tyme White has some great privacy questions about Wordcast and this observation about ad campaigns:

You know what is going to happen, right? Ad campaigns usually run monthly, sometimes quarterly. What if the listener does not listen to your podcast in the month the download it in? I have a ton of podcasts on my hard drive I have not listened to, but I have every intention of doing so. The podcasts that are not time-sensitive (for example interviews) can be listened to anytime and I save them for travel. There can be logical explanations for why there is a discrepancy between when a podcast is downloaded and when it is listened to but I seriously doubt the advertiser will care. This is why they want the information – so they can screw you over and get out of paying for ads that have not been heard.

On personal information: The Audible system doesn’t collect information about the user, it collects aggregate information about the programs played and only on files that require registration or carry audited advertising. So, the data sent back does not say “Mitch listened to time x to time y of program name,” it is “Program name was listened to from time x to time y.” That data is collected for each program without any identifying information, so there is no record of what you listened to. There is, of course, a list of what you’ve subscribed to or purchased used to maintain a personal library.

Targeting-to advertising will be based on the ability to survey listeners at registration. A podcaster could ask for individual information, like an age range or zip code or income, and use that to target ads, but it is at their option.

So, the answers to your questions are:

1.) Will Audible publicly state that they are tracking their users prior to the user downloading the podcast? They aren’t tracking individual user listening, so no.

2.) Will users be able to customize the phone home behavior (as in turn it off)? Will they modify their Privacy Policy to acknowledge the information that is being collected, what is being done with it, how long it is kept and how to opt-out? I’m not sure that the privacy policy needs to be changed, but the Wordcast privacy policy explains that individual user data will not be collected about listening. I have not seen our privacy policy about data collected by the podcaster through surveys at registration for a targeted program, but I’ll make sure there is an opt-out. That option may involve not being able to receive the program if the podcaster is running a controlled-circulation program that bases its ad rate card on qualifying listeners.

3.) Will users be able to customize the phone home behavior (as in turn it off)? No, but remember that not all .aa files report home, only those that are secured or have audited advertising. My podcasts, which are available in .aa format, don’t report home.

4.) Will podcasters be able to download and analyze their statistics? Yes, on a program level that includes what parts of the show are listened to by the percentage of the audience, giving insight into what is or is not working. They don’t get information about what individuals listen to, because it isn’t collected.

5.) What happens if a podcasters podcast downloads greatly differ from the amount of times the ad is listened to? The data will be available to the podcaster and, if they choose to provide that data to the advertiser, to the advertiser, too. This also allows the podcaster to do several things that save money on their end. First, it prevents paying for multiple downloads of the same file by the same user (the cost of unlimited downloads is included in the fee, so users could come back and back and back without the fee increasing). Second, it lets podcasters weed out subscriber files so that if someone stops downloading they can be removed from the subscriber list (manually at first, but the aim is to do it automatically).

There is no system that “red flags” programs based on their ad performance. “Auditing” of listenership means that advertisers can ask for and learn about how much exposure ads actually get, then decide whether to continue advertising. That’s accountability in advertising. So, it’s up to the podcaster to make intelligent decisions about how to include ads so that listeners remain engaged throughout a program. I believe someone who packs their show with ads will alienate listeners.

6.) If the podcaster does not want to partner with Audible any longer will the files be removed from Audible’s servers? Audible would preserve the podcaster’s account for some period of time (I don’t know how long off hand), then delete the data. The data would not be used for any other purpose during that time.

7.) Will there be any restrictions placed on the podcaster – for example getting their own advertisements for show distributed by Audible? I don’t entirely understand your question. The podcaster can sell their own ads and upload them to the system for insertion and we are looking at providing ad inventory, but the how and where of that inventory hasn’t been settled yet.

Technorati Tags: , , , , ,


1 reply on “Privacy, accountability and podcasting”

Tyme has more questions, which took a few days to answer definitively:

1) When the podcast is downloaded it is not given a unique ID? For instance if I purchase something from Audible, the company “should” automatically record the my name, address, phone number, payment information, IP address and of course what I purchased. Does Audible’s system record only that I purchase the podcast or does it associate an ID to the specific podcast? I would think it would for copy protection.

If it is a copy-protected file, yes it will have a unique ID that persists and is related to the user’s account since each file is cooked on the fly, but this is not used for recording individual listening within the file. The data is collected in the aggregate and delivered to the podcaster, but not individual listening.

2) Is there a way to distinguish which .aa files phone home and which ones do not? Podcast listeners are not accustomed to this behavior and it might not even cross their minds this is happening.

Yes, an unsecured file just plays. A targeted or secured file requires you activate an Audible player in iTunes, etc., that would be a free, one-time process in ad-supported file.

3) Here is the key question – when the podcast phones home is the IP address logged in any way? If it is Audible does have the capability to associate the podcast with the person that purchased it because the IP was most likely captured in #1 – if only to help with credit card fraud prevention.

Audible doesn’t log the IP address for untargeted files and while IP addresses can be used to target ads on a geographic basis, the IP address isn’t retained for advertising auditing purposes. If you purchase a program through Wordcast, your IP address would be logged and retained during the transaction, but not when you download the files you bought.

4) In my last question I wanted to know was there a restriction for the podcast to have the same file in multiple locations? For example, hosted with Audible in .aa format and hosted in .mp3 format for listeners that prefer the traditional style.

No, none whatsoever. The podcaster is buying a service from Audible and if they want to offer unaudited MP3s or use BitPass or whatever, that’s up to them. We think this is a great tool for a broad set of problems.

With #3 I am going to be honest, I do not see how the ad can be served without logging the IP address it was served to. As an example of what I am talking about is Sony’s player – it phones home with the ID of the CD that is being played. The difference here is that the customer purchased the CD and did not download it so there is no real way to identify the person playing the CD unless they interacted in some way with Sony’s site.
In Audible’s case the customer is downloading the file so of course Audible has the IP. If the podcast is embedded with an ID then it would be very easy to connect the dots. Now, whether Audible actually would is a different scenario. I am looking for an honest response. In fairness, there are companies that acquire much more personal information (for example Google) and it is not abused. I think people should be informed of the fact then they can make an informed decision.

There’s no question in my mind that you were being honest by just asking. An IP address is logged when you access a Web page—it just happened when you hit this page (which is on my server, under the desk in my office rather than at Audible)—but the question is what happens to the log afterwards. The Sony CDs surreptiously installed a rootkit to enable the communication. Audible doesn’t use the IP address for any purpose of identification of the user and would not retain the records in a way that would be of use to a government if it subpeonaed the records to pursue people.
I absolutely agree that people should be in control of their personal data and so do the folks at Audible. The principle that should drive all data collection is that the user owns their data and companies only borrow it and must keep it in trust. The data Audible collects is strictly controlled and Audible’s site does not link IP addresses to personally identifiable information, such as the programs you listened to (see the Audible privacy policy here). As we continue to evolve the Wordcast platform this is a key priority when weighing changes to the system and data management practices.