When you submit to the Amazon Appstore, the Developer Console asks you to complete a User Data Privacy questionnaire: seven categories, each asking whether the data is collected and whether it is shared with a third party.
You answer it as the publisher, so you need to know what the app we built actually does. Here it is.
Amazon's question | Collected? | Shared with a third party? |
Device or other ID | No | — |
Advertising ID | No | — |
Approximate location | No | — |
Crash logs | No | — |
Diagnostics | No | — |
Other app performance data | No | — |
App interactions | No | — |
The app requests two Android permissions, INTERNET and ACCESS_NETWORK_STATE, and nothing else. There is no analytics library in it, no crash reporting service, no advertising or attribution SDK, and no code that reads a device identifier. It fetches your catalog, plays your videos, and loads your images. That is the whole of its network activity.
Where you left off in a video is remembered, but that is stored on the television itself and never sent anywhere.
We check this against the actual app file rather than from memory. Every Fire TV app we have distributed has been scanned, and all of them answer the table above.
Two things that can change these answers for your app
Read these before you submit. For most customers neither applies, but if one does, the table above is wrong for you.
1. If your app has paid or restricted content
Apps where viewers must activate the device to watch something work differently. The television generates a random ID for itself and sends it to us so we can remember what that device is allowed to play. It is not a hardware identifier and it is not tied to a person, but it is an identifier and it does leave the device.
If your app has an activation screen, answer "Device or other ID: Yes", shared with Tappla as your service provider.
If your app plays everything to everyone, this does not apply and the answer stays No.
2. Where your videos are hosted
Your videos do not come from us. They come from whoever hosts them: Vimeo, Bunny, your own CDN, your live-streaming provider. When the television plays a video, it connects to that host directly, the same way a browser would.
That is ordinary web traffic rather than something the app collects, and it does not by itself make any row in the table a Yes. But two things are worth checking in your host's own dashboard before you sign the form:
Does it show you viewer statistics? If your host reports plays, watch time or sessions back to you, then your host is collecting app interaction data, and you should say so.
Does it break those statistics down by country or region? If so, your host is deriving location from IP addresses, and "approximate location" becomes a Yes, shared with that host.
We cannot see inside your hosting account, so this one is yours to check. It takes a minute and it is the difference between an accurate form and a guess.
If Amazon's review asks about location or phone permissions
Occasionally an automated scan reports that the app "uses" location or phone APIs. It does not, and here is the explanation to send back.
Android apps bundle Google's own support libraries, and those libraries contain helper code for many things the app never uses, including location and telephony. That code ships inside every app built this way. It cannot run here: the permissions those APIs require are not declared in the app, so a call would be refused by Android itself.
If a reviewer raises it, reply that the app declares only INTERNET and ACCESS_NETWORK_STATE, and that the flagged classes are unused AndroidX compatibility code. Forward us whatever they wrote and we will answer it with you.
You still need your own privacy policy
Amazon requires a privacy policy URL on the listing, and it has to be yours, describing your channel and your business. Ours does not substitute for it, because you are the publisher and your viewers are your audience.
If your video host collects viewer statistics, that belongs in your policy too.
Roku and Apple TV
Roku and Apple ask a similar set of questions when you publish there, and Apple asks at every single submission for the App Store privacy label.
The apps are built separately for each platform, so we checked each codebase rather than assuming they match. The answers come out the same.
Roku. No advertising identifier: Roku's RIDA is never read. No analytics library, no crash reporting, no location. What the app stores in the Roku registry stays on the television.
Apple TV. No advertising identifier and no vendor identifier. No analytics library, no crash reporting, no location. The project links Apple's CoreLocation framework but never uses it, and carries no location usage-description key, so the app could not obtain a location even if it asked.
The one exception is the same on all three platforms, and it is the activation case above. An app where viewers link their television sends a device identifier when they do. On Fire TV and Apple TV that is a random ID the app generates on first run; on Roku it is the per-channel identifier Roku itself hands out. Neither is a hardware identifier and neither is tied to a person.
Roku and Apple word their forms differently from Amazon. Tell us which one you are filling in and we will map these answers onto its wording.