Another basic request I have: if you are building your own video player, make sure it's not MUCH worse than system's default. This is particularly glaring in Apple TV: system player has really decent UX, and almost all third-party implementations are terrible. Slow, laggy, jumpy or some mix of these. Scrubbing is commonly the worst aspect.
The YouTube app on Apple TV is a giant middle finger from Google to their users.
I’m convinced that not a single person on that team dog foods on an Apple TV at home.
I wouldn’t trust myself the app on an Apple TV around children because I wouldn’t be able to hold back my cursing when attempting to scroll a list, selecting an element, using the seekbar, or use basic functionality implemented on other platforms like “add to watch later”.
The Apple TV YouTube app also manages to break typing into its search field when using the iPhone remote, with letters frequently getting transposed. It’s infuriating.
Contra the domain name, this "spec" strikes me as entirely reasonable. It's remarkable how poor the UX is on apps from some of the world's largest media companies and service providers.
I'd also add that any search should be for metadata attached to that piece of content. If I search for Jackie Chan, I want to see Jackie Chan films (or appearances in TV episodes) vs. the only film with Jackie Chan in the title.
On a similar vein, it seems like most streaming services don't let you browse actor catalogues. Thanks Prime, for showing me a list of unclick-able actor pictures. It seems pretty obvious that someone might want to select an actor in something they just watched and see all the other things that actor is in, especially on something like Prime where you can presumably purchase or rent most if not all of the other things they're in?
I like this spec, and I also like the concept of an unsolicited spec as a format. It's like a Considered Harmful but it actually tells you what to do instead!
Keep selected user profile (don't ask my kid to select their profile every time they open the app [e.g. on a Roku only they use]), disable auto-play, disable video logos on app startup. The latter two should actually be OS level settings that any app must respect.
If I had to add another thing, please for the love of god do not reinvent the video player for the sake of re-inventing and don't cripple it for fun.
I'm looking at you Netflix, who until very recently didn't have playback speed settings but also use HTML5 videos so if you have an extension that allow you to change it but would work for a few seconds and then break the Netflix stream because they are doing something weird with it.
It not a DRM Thing because it works just fine on prime video.
Also like the post said prime video please regroup all of the seasons under a single jacket, telling me to continue watching TV show x at season 1 when I've already binged through to season 3 without clicking on their jackets is a sucky UX.
On Apple TV, this is by far my largest gripe with streaming apps. Consistently, the worst apps on the platform are those that decide they need a bespoke player UI. YouTube is one of the worst.
The stock video player is perfectly fine, please don’t screw with it excessively!
The YouTube app on Apple TV is an excellent example of how to design a user interface designed specifically to frustrate.
Favorite example: you have to press "Menu" to hide the player UI while a video is playing (the native one just requires a tap on the touchpad). It'll also auto-hide after a delay (a longer delay than the native video player, of course), and once it's hidden, pressing "Menu" will instead back you out of the video.
So this leads to my favorite dance, which is:
1. Play a video
2. Notice the player UI isn't going away, and the YouTube player UI is super obnoxious and blocks half the screen, and I want to see the video
2. Press "Menu" to hide the player UI
3. Oops! right before I hit "Menu" the player UI auto-hid, and now YouTube took me back to Home (which, bafflingly, does not remember my video position, so if I just resume on the same video, it'll start at the beginning)
4. Now I have to navigate to Library > History (which _does_ remember my position) and find the video, just to keep watching from where I was.
Countless other complaints of course. The saddest thing is it just isn't getting better -- I can't remember the last functional update, it must have been years ago.
It's a competitive advantage so this probably will never happen, but I wish Google would extract the YouTube player and publish it as an OSS library. As much as I hate the (ever changing) rest of the YouTube UI the player is the best I've used on the web by an insane margin.
I like this a lot and agree with pretty much all of it.
I'd add: Figure out a way to detect that I'm doing a re-watch of a series and do something with the progress bar on episode previews to reflect that. So if I've watched all of a series/season and then watch episodes 1, 2 and 3, maybe the progress bar on episodes 4+ reset themselves to an unwatched state or have some kind of greyed-out tint to show where I left off this time around.
Netflix used to support my Roku's "skip back 10 seconds" remote button, but years ago they stopped. I rarely watch Netflix anymore but just played something the other day and found they have now also dropped support for the system's subtitle framework, meaning I can no longer use the "show subtitles only when I rewind" feature, which I use constantly.
I cannot conceive of any reason for dropping these features which they already implemented except a petulant desire to maintain full control of the user experience rather than ceding anything to the platform conventions. Well, I bought a Roku because I like the platform conventions. Since Netflix seems intent on pissing in my cheerios I'll happily be pirating what little interesting content they put out in the future, so I can watch it in Plex where things work as they should.
It doesn't on mine, it just brings up the timeline, where you then have to press it more times to jog back then press select. Surely you see how that's not the same thing?
I've found the heuristics streaming apps use to determine whether or not to show "play next episode" or "continue watching" to be very hit or miss (sometimes due to a bad heuristic, other times due to a given episode having an unusual end point). Providing an accessible means for the user to quickly course correct when this heuristic inevitably breaks would be handy.
As a concrete example, I frequently find Funimation's app recommends I "continue watching" an episode that I'd consider already completed, and unfortunately the quickest way to "fix" instead of backing back out and browsing around is to frantically scrub through to the end so it auto-plays to the next episode.
Another: Please don't put the scrub (position) bar so near to the other icons that you end up clicking "Next Track" or "Captions On" every time you try to move position in the video.
And: Don't make the scrub bar so small that you have to find a small child to click on it for you as your fat fingers can never select it.
This isn't always true. Some shows release specials that aren't linked to any season, but go between seasons. Some shows (like web serials) release in one continuous stream. Sure there are workarounds/hacky solutions to these, but assuming all shows have the exact same hierarchy is a bad assumption to make.
MyAnimeList has a pretty good navigation list IMO. For example on https://myanimelist.net/anime/25777/Shingeki_no_Kyojin_Seaso... there is prequel, sequel, other (spinoffs and specials), and "alternate version" (condensed movie summary or remake). Maybe the naming could be better but those relationship types are the basic set for any kind of serialized media. Note that the media type is not fixed, e.g. the sequel to the last (TV) episode of Firefly is Serenity (a movie).
Now it's true that some shows aren't really designed to be serialized, e.g. for the Fate series there is a complicated mess of a graph (https://i.redd.it/6fz5fzz6sfv11.png) and most relationships will have to be encoded as "other", but since it's so messy, almost any watch order is reasonable, so there isn't much structure lost.
These ideas extend well beyond the realm of streaming apps too - so much of the software I use day to day is shockingly bad. Far, far too much churn driven from ladder-climbing obsessed middle managers who want rewrites and redesigns and engagement to go up up up.
What drives me nuts these days is how the skip-back in Netflix on TVs /firesticks/ etc will sometimes skip back 10 seconds, and sometimes bring up a scrub bar of images. Selecting one image to the left often re-starts playback at exactly where you hit the button in the first place, and moving one farther to the left often leaves you 20 seconds back. Grrrrr
One mistake that makes me laugh/cry is how Netflix on iPhone asks me "who is watching?" nearly every time I open it. On a laptop or iPad, sure, but on the phone, it could be pretty sure it's still me, right?
I thought this too, but parents hand their kids phones to watch something on all the time, and having a Kids profile is more elegant than hunting in a menu for a Kids Mode.
Clicking or tapping on the background ‘behind’ a modal dialog should close that dialog. Netflix and Hulu keep breaking and fixing and breaking and fixing this.
Some clicking / tapping controls should be available in the video player without first clicking / tapping to make the controls appear. C/t in the center should play / pause. Generally on the left and right sides should go back / forward by n seconds.
After a commercial break, seeking back to re-watch what happened before the break, or back to catch what you missed right after the break, should not force re-watching of the commercials. Hulu is horrible about this.
Edit: Also, allow seeking backwards in commercials!! Some commercials are interesting in some way. Why on earth would you prevent customers from watching more commercial time?
This is a great list, and OP's frustrations with existing services match mine quite well.
> On launch, it must be immediately obvious how to resume watching whatever the user was watching previously. This may be the most important feature outside the video player itself.
> If the user was in the middle of watching an episode of a TV show, the most prominent thing on the screen should be a way to continue that episode. If the user just finished an episode, then “resuming” means watching the next episode, and so on.
With my speculate hat on, I've hypothesized the lack of this screamingly obvious feature is a deliberate choice designed to encourage users to "discover" new shows, since I imagine that's a metric they want to increase.
When I worked in the industry this was part of the reason, but another part was "data driven" AB testing on "what results in higher watch minutes per session" and similar numbers - one hypothesis was that there's a significant cohort of users who stopped watching something because they didn't like it and so for them it's wasting real estate that could be used for finding something new.
I put "data driven" in scare quotes there because you can observe the immediate per-session metrics but you can't observe the annoyance or long-term frustration with the platform. Or, say, long-term fall-off of watch behavior because now you finished ONE of the things you had had in progress but forgot about the other series because the "continue watching" row isn't presented up front, you had to go dig every time.
If you're interested in these challenges (and trust me, these things all seem simpler than they are in practice), consider coming and joining us at Philo (https://about.philo.com/jobs/)! We're hiring for basically all technical (and plenty of non-technical) roles/specialties/platforms, so if you're excited about dragging TV kicking and screaming into streaming age, drop me an email: andrew@philo.com
The same general principle applies to watch Netflix as shopping at Amazon: they do not really care what you watch, as you as you watch something. If that’s continuing from before, that’s fine. If it’s something they’re promoting, or a new show they can entice you with, that’s great too. Maybe even better — now there’s two shows you’ll want to finish.
I mean, sure, they’d probably prefer you watch an original property they have more control over than something with a costly license, but overall the primary concern is keeping usage levels up and cancellations down.
They could recruit more users to their paid version, Youtube Premium. Also, their sister app Youtube Music does compete with other audio apps like Spotify. They have the advantage of showing videos, but that advantage is heavily marred by the terrible UI of their player.
It’s the curse of cross platform development. Apple TV has a tiny user base. It’s not economical to do much custom development for it. So streaming apps tend to treat it as another “smart tv”, which means very limited features.
I wonder how much of this is a lack of competition around the apps themselves. I don't think people subscribe to Netflix or Disney+ primarily based on what the app experience is, I think that as long as the app isn't horribly broken they probably subscribe based on the exclusive content and they put up with some extra friction.
With current DRM laws this is impossible to have as a market, but hypothetically if anyone could build a player for Netflix, I wonder if Netflix's official app might not start to improve. Even more impossible scenario, but hypothetically if both Netflix and the Disney+ catalog were available from both services, the only distinguishing feature between those services would be which one had a better app and user services and recommendations and features around that content.
At least with Youtube I can use unofficial apps like NewPipe that have much better UI experiences than the official apps. Even that isn't perfect, it's hard to call NewPipe a "competitor" for Youtube's official client on Android. But the situation is still a lot better than being locked into an experience that is allowed to be bad because there's no reason for it to be good. I mean, if you're on Disney+ because your kids want to watch classic movies, how bad can the client get before you actually decide to switch? The bar to keep you on the service is probably not that high.
Maybe I'm over-cynical, but I'm somewhat surprised occasionally that these kinds of apps are even as good as they are given that (as far as I can tell) the primary competition between services is mostly just content.
One that annoys me a lot is the video controls overlay auto hide timer. I'm never sure if the app hasn't one, or if I'm about to hit back at the wrong time and stop the video when I just try to hide the overlay. If you put a timer, always add a visual indicator too (reverse progress bar in a corner for example).
Looks like a lot of people here agree with this and have similar frustrations which makes me wonder - surely some of the developers working on these apps dogfood their own work, if not for any other reason than it's just very popular. The presence of these issues means that they can't fix them. Why?
I mean, I've seen this happen where developers legitimately want to fix little annoyances in the UI, but they can't. I think the reason why is a combination of two things. First, a mature product has to handle a ton of different scenarios making it a major PITA to change anything. If it's not a quick fix, it has to go trough the normal product process which brings me to the second reason - the product managers don't get evaluated based on how usable something is, rather it's some sort of an engagement metric. As long as the watched minutes go up, nobody cares if people are throwing their remotes in rage. So this is on product leadership, but maybe they just need better metrics, like a rage button on the remote.
49 comments
[ 8.0 ms ] story [ 103 ms ] threadI’m convinced that not a single person on that team dog foods on an Apple TV at home.
I wouldn’t trust myself the app on an Apple TV around children because I wouldn’t be able to hold back my cursing when attempting to scroll a list, selecting an element, using the seekbar, or use basic functionality implemented on other platforms like “add to watch later”.
Or "Falsehoods programmers believe about streaming apps"
I'm looking at you Netflix, who until very recently didn't have playback speed settings but also use HTML5 videos so if you have an extension that allow you to change it but would work for a few seconds and then break the Netflix stream because they are doing something weird with it. It not a DRM Thing because it works just fine on prime video.
Also like the post said prime video please regroup all of the seasons under a single jacket, telling me to continue watching TV show x at season 1 when I've already binged through to season 3 without clicking on their jackets is a sucky UX.
The stock video player is perfectly fine, please don’t screw with it excessively!
Favorite example: you have to press "Menu" to hide the player UI while a video is playing (the native one just requires a tap on the touchpad). It'll also auto-hide after a delay (a longer delay than the native video player, of course), and once it's hidden, pressing "Menu" will instead back you out of the video.
So this leads to my favorite dance, which is:
1. Play a video
2. Notice the player UI isn't going away, and the YouTube player UI is super obnoxious and blocks half the screen, and I want to see the video
2. Press "Menu" to hide the player UI
3. Oops! right before I hit "Menu" the player UI auto-hid, and now YouTube took me back to Home (which, bafflingly, does not remember my video position, so if I just resume on the same video, it'll start at the beginning)
4. Now I have to navigate to Library > History (which _does_ remember my position) and find the video, just to keep watching from where I was.
Countless other complaints of course. The saddest thing is it just isn't getting better -- I can't remember the last functional update, it must have been years ago.
I'd add: Figure out a way to detect that I'm doing a re-watch of a series and do something with the progress bar on episode previews to reflect that. So if I've watched all of a series/season and then watch episodes 1, 2 and 3, maybe the progress bar on episodes 4+ reset themselves to an unwatched state or have some kind of greyed-out tint to show where I left off this time around.
I cannot conceive of any reason for dropping these features which they already implemented except a petulant desire to maintain full control of the user experience rather than ceding anything to the platform conventions. Well, I bought a Roku because I like the platform conventions. Since Netflix seems intent on pissing in my cheerios I'll happily be pirating what little interesting content they put out in the future, so I can watch it in Plex where things work as they should.
> Skip to the beginning or the end of the video.
What critical use does skipping to the end of a video have?
As a concrete example, I frequently find Funimation's app recommends I "continue watching" an episode that I'd consider already completed, and unfortunately the quickest way to "fix" instead of backing back out and browsing around is to frantically scrub through to the end so it auto-plays to the next episode.
And: Don't make the scrub bar so small that you have to find a small child to click on it for you as your fat fingers can never select it.
This isn't always true. Some shows release specials that aren't linked to any season, but go between seasons. Some shows (like web serials) release in one continuous stream. Sure there are workarounds/hacky solutions to these, but assuming all shows have the exact same hierarchy is a bad assumption to make.
Now it's true that some shows aren't really designed to be serialized, e.g. for the Fate series there is a complicated mess of a graph (https://i.redd.it/6fz5fzz6sfv11.png) and most relationships will have to be encoded as "other", but since it's so messy, almost any watch order is reasonable, so there isn't much structure lost.
Some clicking / tapping controls should be available in the video player without first clicking / tapping to make the controls appear. C/t in the center should play / pause. Generally on the left and right sides should go back / forward by n seconds.
After a commercial break, seeking back to re-watch what happened before the break, or back to catch what you missed right after the break, should not force re-watching of the commercials. Hulu is horrible about this.
Edit: Also, allow seeking backwards in commercials!! Some commercials are interesting in some way. Why on earth would you prevent customers from watching more commercial time?
> On launch, it must be immediately obvious how to resume watching whatever the user was watching previously. This may be the most important feature outside the video player itself.
> If the user was in the middle of watching an episode of a TV show, the most prominent thing on the screen should be a way to continue that episode. If the user just finished an episode, then “resuming” means watching the next episode, and so on.
With my speculate hat on, I've hypothesized the lack of this screamingly obvious feature is a deliberate choice designed to encourage users to "discover" new shows, since I imagine that's a metric they want to increase.
I put "data driven" in scare quotes there because you can observe the immediate per-session metrics but you can't observe the annoyance or long-term frustration with the platform. Or, say, long-term fall-off of watch behavior because now you finished ONE of the things you had had in progress but forgot about the other series because the "continue watching" row isn't presented up front, you had to go dig every time.
I mean, sure, they’d probably prefer you watch an original property they have more control over than something with a costly license, but overall the primary concern is keeping usage levels up and cancellations down.
With current DRM laws this is impossible to have as a market, but hypothetically if anyone could build a player for Netflix, I wonder if Netflix's official app might not start to improve. Even more impossible scenario, but hypothetically if both Netflix and the Disney+ catalog were available from both services, the only distinguishing feature between those services would be which one had a better app and user services and recommendations and features around that content.
At least with Youtube I can use unofficial apps like NewPipe that have much better UI experiences than the official apps. Even that isn't perfect, it's hard to call NewPipe a "competitor" for Youtube's official client on Android. But the situation is still a lot better than being locked into an experience that is allowed to be bad because there's no reason for it to be good. I mean, if you're on Disney+ because your kids want to watch classic movies, how bad can the client get before you actually decide to switch? The bar to keep you on the service is probably not that high.
Maybe I'm over-cynical, but I'm somewhat surprised occasionally that these kinds of apps are even as good as they are given that (as far as I can tell) the primary competition between services is mostly just content.
I mean, I've seen this happen where developers legitimately want to fix little annoyances in the UI, but they can't. I think the reason why is a combination of two things. First, a mature product has to handle a ton of different scenarios making it a major PITA to change anything. If it's not a quick fix, it has to go trough the normal product process which brings me to the second reason - the product managers don't get evaluated based on how usable something is, rather it's some sort of an engagement metric. As long as the watched minutes go up, nobody cares if people are throwing their remotes in rage. So this is on product leadership, but maybe they just need better metrics, like a rage button on the remote.