Hacker News Reader

Comment by sidlls | Hacker News Reader

If I were to seriously attempt to commercialize the toy apps I build for my kids/family I'd go iOS first easily. People with Apple devices are far likelier to pay, and Android development is, in my view, an inferior experience (although Kotlin makes it marginally less so these days).

View on HN

This may have been historically true, but our internal data shows us that starting early 2017, people in tier 1 countries were equally likely to pay and have comparable LTV on both platforms.

And if you also going to launch in EU/Australia/Canada, you will have a nearly 50-50 split of ios and android users on a platform neutral app.

So the classical logic of implementing on iOS first doesn't really apply anymore. Now its more, implement first on the platform you are comfortable with but don't forget about the other platform because there is a equal amount of money to be made there...

PS: Data excludes low end iOS and android devices. (< SE, old iPads/iPods, old and cheap android devices). These devices have extremely low conversion on both platforms anyway.

The logic still applies.

1. Start with platform you're most comfortable with. 2. Iterate on product until you have a winning formula 3. Replicate on 2nd platform to double revenues

Building is a lot easier than identifying what to build, better to speedily experiment on one platform.

In my opinion the new logic is build in React Native (unless your app fits in the 10% of use cases that won't work well with RN), then deploy to both at once. We have had excellent results.

My take is that RN is great for demo apps or simple UIs, but I would not choose it for a production app where performance matters (ie. using camera, audio, custom controls, platform-specific SDKs, multi-threading, etc). You're definitely going to do heavy lifting in native code, and a hybrid app quickly becomes a mess. Dealing with multiple environments and super slow build times in Xcode will seriously slow down development cycles.

IMO RN is no exception to the Simple vs Easy trade-off [1] Not to mention using RN gives FB a legal blank check from your company [2]

[1] http://chrisfrost.com/wp-content/uploads/SimpleVsEasy-478x51...

[2] https://medium.com/@raulk/if-youre-a-startup-you-should-not-...

That article is out of date. As of September 2017, FB has re-licensed React under the MIT open source license and dropped the patent stuff.

I don't think you contradicted anything I said. I would say special camera apps, multi-threading etc. are not needed that often.

As for slower build times, that is the beauty of RN. Building is not needed most of the time just an initial build and a rebuild every time some of the base configurations change or you are creating code outside of RN.

> fb dropped patent stuff

Ah, thanks for the update, glad they did that

> are not needed that often

Right - statistically most apps are probably throwaway code for hackathons or demos. However, I'd venture to say that >50% of serious apps will end up needing performance boosts from native code.

> Building is not needed most of the time

You need to rebuild every time you make a change in native code (obj-c or swift). Hot reloading is cool for JS changes, but bloat from RN becomes a nightmare when debugging obj-c or swift issues.

React Native is great for apps, and if it's a game, just use Unity. They're building up their internal CI infrastructure and it works better every month.

Tbh, I'm not sure how many use cases don't fit either React Native or Unity, I'd practically never choose to do an app another way these days.

I rather use native UI code + C++ for business logic, use Qt or Xamarin than even bother with React Native.

I'd say RN won't work well in 99% cases.

I'd say you are probably using a lot more React Native apps than you realize right now.

Could you elaborate?

I wish everyone downvoting you would at least state their problem with your post

That's interesting. I am admittedly using my recollection of data I saw a year or so back. Has the fraction of iOS users who pay gone down or the fraction of Android payers gone up?

Conversion on android has gone up, along with users on high-end android devices. At the same time CPI on iOS has gone up.

Speculation: iOS users may be running into app fatigue and just don't want to install new apps anymore.

Have you thought about writing up your findings? I haven't seen any report break out the data like that before.

Is this referring to free apps, paid apps, or both? I've had a paid app on both stores for about six years now and in my experience the iOS version makes about 10x the Android version every year. Many other developers have reported a similar discrepancy.

Sorry should have clarified, I am taking about freemium apps , both subscriptions and IAPs. Don't have data on paid apps.

Also I am talking about app that are well monetized and advertised so you have an level comparison instead of chart positions giving and sustaining a significant lead on one platform just by some fluke.

Edit: s/comparable LTV/comparable ROI(~= LTV-CPI)

You will have equal ROI, with the LTV being slightly lower on android but so too will the acquisition cost... overall profit being the same.

But would you really setup a CI process to automatically build that app? (and one that'll probably cost you at least a 100$ a month)

I would. Every time. I've been happy with bitrise if people are looking for mobile CI alternatives: https://www.bitrise.io/

Bitrise looks very nice for mobile apps. And it is open source https://www.bitrise.io/cli Are people interested in if we ingrate https://github.com/bitrise-io/bitrise in GitLab Runner so it works with GitLab CI?

We are happy to discuss it in more detail! DM me on twitter @birmacher

Awesome, I just followed you from https://twitter.com/sytses and I think I need a follow back before we can DM

Bitrise has been great for us at work as well, but we're beginning to investigate Microsoft's CI platform since we build web/desktop apps as well, which aren't a first class citizen in Bitrise.

Fwiw: We also wish there was a way to publish test/linting results (you can make it an artifact, but it won't load assets, so you're stuck zipping and downloading or publishing to your own managed Web server)

Reply on news.ycombinator.com