Re: [w3c/manifest] Allowing only site-triggered install prompt (#627)

Hi, I thought to share my opinion regarding this topic, I share a lot of points of @photopea.
I appreciate @marcoscaceres replies, you gave me more insight about the problems spec/browser are having. I hope this post will help you and if there is anything I can do I am available to help.

I will start with the problems and then looking for some solutions.

All this points comes out when I started to integrate PWA myself, are problem I had to discuss with business and designers.

## Problems

1. **Unpredictable install action** You have an event to listen when the banner comes up. You can even postpone the banner. If I have a page where I explain my service and I have a section where I place an install button, I need to create a design which takes into account when I can trigger the banner and when I cannot. One where you can click a button and it will open web app install banner and another one where you need to train the user to install the web app through the browser UI (setting -> add to homescreen). In the later case I would need to create at least 6 screens to explain how to do it because we have two main mobile browser and 4 desktop browsers.

2. *When you uninstall you cannot install it again* This problem I think is quite serious here. Website are visited and dismissed frequently so I guess is common the following case: 
you go to a website you think is a good service, you add to homescreen then after a some days you realise you are not using it that often and you remove it. But people change their needs and habit maybe in 4 months they will need that service again and they could change their mind and become a regular user. A product is never finish, you add new feature, improve, change products but next time they visit they will not have a way to install, you will have to train them how to use the browser and click on setting and find the menu item add to homescreen.

3. "browser inconsistency" I already see we are going toward that direction. Any browser is using a different approach. Chrome have a banner, Edge use the window store, Safari you still need to use the share button, Firefox add to homescreen badge. It makes it hard for developer and it is even more problematic because we cannot feature detect these behaviours.

4. **inconsistent permission UI**. I consider add to homescreen a permission screen. It's hard for a user to keep track of all differences browser UI. We should be more consistent, even because user already think popup as virus or dangerous thing to close. If you give them a look which is similar they will be little bit less worry about them.

## Possible solutions
1. **block add to homescreen after two try in the same day** I think add to homescreen popup should not display immediately by default. User can trigger the install permission only when they click on a UI element which is not on position fixed/absolute or even if that element has parent has those attributes. it should be a genuine user interaction.

2. **web app === app** User should not bother about technical details. Browser should communicate correctly to user. For instance if a website have manifest which use display standalone should be consider an app if it does not should be consider a link because it opens with the browser shell.
web app => open independently
website => open within browser

3. **badge increase noise** It will sound contradictory here I kind of like ambient badges but I am worry we are going to confuse user even more:
<img width="372" alt="screen shot 2018-04-11 at 01 57 10" src="https://user-images.githubusercontent.com/4461152/38590908-03a1ddde-3d2c-11e8-8560-93df4fda039a.png">
<img width="514" alt="screen shot 2018-04-11 at 01 57 21" src="https://user-images.githubusercontent.com/4461152/38590910-06502d06-3d2c-11e8-97d9-0333656cc826.png">
When I tested this feature with some users they thought it was going to add the website to the browser homescreen not the phone homescreen. Also we have two home icon very similar.
I think google drive got this better.
<img width="237" alt="screen shot 2018-04-11 at 02 01 05" src="https://user-images.githubusercontent.com/4461152/38590985-61b00b30-3d2c-11e8-8fe9-01eac413a235.png">
If we leave the website to explain why they need to add to homescreen they will do a more contextualised job than a browser fit all solution. I do not think ambient badges are something user will use often also. But if we cannot have better control on install action this is probably the best third option but it needs to be consistent across browser.

4. **app store** windows took an interesting move when they added PWA in the window store. We trained user for years that apps are inside app store. It is intuitive for user to keep doing that. If you ask a user to install an app where they think they will go. I am sure they will search first on the app store. This option would be my second best option for installing PWA but I guess Apple will never agree on this so probably better rely on a more consistent option.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/w3c/manifest/issues/627#issuecomment-380295542