Firebase at Google I/O
I’m sure a lot of you are aware of or even attended Google I/O earlier this month. In Liverpool, we hosted an I/O extended event to live stream the keynote, and it was great to see so many local developers and enthusiasts turn up to watch it with us, even with Liverpool playing in the Champions League that night!
A common topic at the conference was of course Firebase. With 14 different talks concerning it, it’s safe to say there was a variety of new features to learn about. I’ve spent the last few weeks catching up with these talks, and I thought I would share what I’ve found.
What’s New in Firebase
An obvious choice, but a good one. This talk highlights all the new and sparkly features Firebase is bringing to us developers in 2019.
This talk combines live demos, case studies, and code examples, and is a great way to quickly understand what to try out next, and how they might help you as a developer.
A particular highlight for me is AutoML. I’ve always found Firebase ML really interesting but felt it inaccessible as I have no idea how to start without a pre-populated data set. It’s really exciting to see Firebase come up with ways to make Machine Learning even more accessible, and opens up the possibilities for everyone to make the most of it in their apps.
Zero to App: Live Coding a Cross-Platform App on Firebase
As an iOS and Android developer, I’m always a bit of a sucker for any sessions that focus on cross-platform development. I love being able to see features side by side on both platforms, and it’s a really good way of bringing mobile developers closer together.
I particularly enjoyed the style of this session, as it was all about collaboration and communication, while also showcasing the best features of Firebase.
ML Kit: Machine Learning for Mobile with Firebase
After watching the session focusing on new features on Firebase, and seeing the updates to MLKit, this ‘ML for Mobile’ session was of particular interest.
The way this presentation was mixed with live demos and code samples was really interesting and is a very engaging talk.
Some particular highlights for me were the offline language processing, which was shown to be incredibly fast for translations, and also the collaboration with IKEA, which lets you find matching or similar items in the IKEA catalogue.
As someone with a creative background, the collaboration with the Material Design team was really exciting. Sometimes more technical features can ignore the user experience for developers, and it’s fantastic to see this being thought about, while also making it easier to make beautiful apps while utilising MLKit.
Finally, the introduction of custom models is what had initially attracted my attention from the main session. I particularly loved the example of identifying dog breeds!
The user interface for the AutoML tab, which guides you through creating your own data sets, has really helped me understand more about the data and model that I want, whilst also taking away a lot of the complexity that has blocked me from approaching machine learning in the past. Being able to pick and choose between different sizes and training times makes it feel far more flexible.
Overall this is probably my favourite talk. I found the presenters were all really clear, and the live demos really helped me understand how I might use these new machine learning features myself.
Understand and Engage Your Top Users with Dynamic Audiences
Analytics is something that as an app developer, is really important to understand and utilise to the maximum of their potential. Having overall project level analytics is incredibly useful, and as a cross-platform developer, it is invaluable.
Enhanced filters
This new feature adds a huge amount of flexibility to understand the data provided by Analytics. The example they gave was whether completing the tutorial in a game meant that you stayed playing the game for longer and gave you better scores. Being able to apply these filters across platforms enables you to understand overall trends that might be less platform specific.
Being able to utilise Audiences to compare and contrast the data between different groups allows you to really quickly understand any differences. The removal of limitations on what and how many filters you can apply means that you can dive incredibly deeply into your data, which can only help you improve your own apps.
The particular highlight of this is the ability to make Audiences based on “Observed Behaviour”, which you can then use to target these users using Cloud Messaging, Remote Config and Google Ads. This means you can potentially address any issues or user behaviours immediately without requiring a new app build.
Furthermore, the Audience Builder interface is fantastically easy to use and is really clear on what data you will be collecting and what events you want to target by including the Summary card that gives visual guides for your data sets. I also loved the recommended analytics events that Firebase thinks you should use depending on your app’s category.
Analysis
The announcement of the new feature “Analysis” was really exciting to me. This is accessible through the Analytics dashboard once you’ve linked to Firebase. This gives you incredibly detailed access to your data, including visualisations such as Exploration, Segment Overlap and most interestingly for me, Funnel Analysis.
The ease of getting started with this immediately from Google Analytics is brilliant, and the UI is immediately easy to understand, which is something I’ve found difficult in the past with data platforms. The visualisation of creating a funnel completely makes sense, and you can get previews of how your data might look, and get immediate feedback on your app with barely any setup. The ability to “View Abandonments” is particularly interesting. It allows you to understand deeply your users who drop out of your defined funnels, which could prove incredibly valuable!
The period comparison feature within the scatter plot graph was really interesting, and it is so easy to understand the trends in your data and identify and declines before they become serious problems. You can then link this in to the Audiences features to target any issue groups, giving you huge control over your app and your users.
Analysis rolls out this Summer.
I’ve only scratched the surface in this post on the range of topics and things to learn about Firebase from I/O. I’d really recommend watching the talks I’ve highlighted above, and you can find the Firebase at I/O playlist here for the rest.
