Creating an Android app in 2026 still follows the same arc: have an idea, validate it, design it, build it, test it, submit it, and maintain it. What's changed is that every stage now has a gate you can't skip. Android 16 brought Material 3 Expressive, Google Play won't accept new apps that don't target API 36 from 31 August 2026, developer identity verification is now mandatory (even for apps you distribute outside Play), and on-device AI is a free, built-in feature. And the smartest early decision for most UK businesses isn't "Android or not", it's whether to build for Android only or for Android and iOS at once.
I'm Gareth, CTO at Foresight Mobile. The 2023 version of this post was a thin overview that spent half its length on app builders and freelancers. Android development has moved on enough that it needed a proper rewrite. Here's how to actually create an Android app in 2026, stage by stage, with the decisions that matter at each one.
Android still runs roughly 70% of the world's smartphones, so for global reach it's the bigger platform by a distance. The UK is the exception worth knowing about: here it's close to a coin-flip, around 52% iPhone to 47% Android, one of the few markets where iOS edges ahead. That matters for your build decision, because reaching the whole UK audience means both stores, not just one.
It matters more than ever because the app economy now runs on spend, not downloads. Global consumer spending on apps hit $155.8 billion in 2025, up nearly 22% in a year, even as total downloads fell. People are selective and they pay for quality. A well-built Android app earns its place; a rushed one gets uninstalled. With that framing, here's the process.

Every app starts with an idea, but in 2026 the cheapest way to get it wrong is to start building before you've checked anyone wants it. Validate first. Confirm the audience, the problem and the willingness to pay before you commit a build budget. Our App Gameplan does exactly this: a fixed-price, four-week discovery that gives you a board-ready answer on whether to build and what to build, with the fee credited if you go ahead. It's the cheapest insurance you'll buy on the whole project.
Android had its biggest design refresh in years. Android 16 shipped in June 2025, and alongside it Google introduced Material 3 Expressive: a new spring-based motion system, bolder colour, background blur for depth, fresh icon shapes and refreshed typography. It started reaching Pixel devices that autumn and is now the look users expect. Android 17 arrived in June 2026 as the current release, so the platform moves at an annual pace you need to design with, not against.
The practical point is the same one iOS designers face with Liquid Glass: a UI built for flat, older Material chrome looks dated now. Build against the current Material 3 components and you get the Expressive motion and colour treatment more or less for free, so designing on the current toolchain is the cheap path to an app that looks native rather than two years behind.
Here's the decision the old version of this post missed entirely. The real choice in 2026 isn't "app builder, freelancer or agency". It's whether to build for Android only, or build for Android and iOS together from one codebase.

Native Android gives you the exact Android 16 look and feel, the deepest platform integration, and the newest APIs the day they ship. It's the right call when Android genuinely is your whole audience, when the brief is "must feel like a flagship Android app", or when you're pushing performance at the metal. That's our native Android development service, built with Kotlin and Jetpack Compose. But in the UK, where the market splits roughly in half, building Android only leaves half your audience on the table, and building twice is the most expensive way to reach both. A cross-platform codebase lets one team and one budget cover both stores with a single roadmap, which is our cross-platform app development service. We build cross-platform with Flutter, and native with Kotlin and Compose when the brief calls for it. The honest trade-off: native wins on flagship feel and bleeding-edge platform APIs; cross-platform wins on cost, time-to-market and reach. For most UK businesses that want both stores, cross-platform wins.
For native Android, the 2026 toolchain is Kotlin 2.3 with Jetpack Compose in Android Studio. Compose is now the mainstream, declarative way to build Android UI, the same shift SwiftUI brought to iOS. The genuinely new capability worth designing in is on-device AI. Google's Gemini Nano model runs inside the Android AICore system service, and you reach it through the ML Kit GenAI APIs: high-level interfaces for summarisation, proofreading, rewriting, image description and structured text generation, with no server, no per-call bill, and data kept on the device. AICore isolates each request and keeps no record of inputs or outputs after processing, so privacy is the selling point.
One honest caveat. On-device generative AI runs only on capable hardware, the recent flagship MediaTek, Qualcomm and Google Tensor chips, so it isn't universal across every Android phone. Plan a graceful fallback for older devices. Building these features properly, with that fallback thought through, is what our AI app development service focuses on.
Test thoroughly, then clear the gate that catches a lot of teams: the target API level. From 31 August 2026, new apps and updates submitted to Google Play must target Android 16 (API level 36). Miss it and you can't submit, and your app becomes invisible to new users on the newest devices. This isn't a one-line version bump either. Android 16 ships behaviour changes that activate when you target it, such as stricter predictive-back enforcement, so existing apps need a real test pass rather than a number change. There's a fuller walkthrough in our Android app publishing guide.
The Google Play Console costs a one-time $25 registration fee, paid once rather than annually, which is a genuine point in Android's favour if you're weighing both stores against Apple's yearly charge. Apps now ship as Android App Bundles rather than raw APK files.
Two 2026 gates surprise first-time publishers. First, if you're on a new personal Play Console account (created after late 2023), you must run a closed test with at least 12 testers, opted in continuously for 14 days, before you can apply for production access. Organisation accounts and older personal accounts are exempt, which is one practical reason to publish through an established account or agency. Second, Google now requires developer identity verification to keep apps installable on certified Android devices, and this reaches apps distributed outside Play too, including third-party stores and direct downloads. Verification opened to all developers in early 2026, and enforcement begins in its first wave of regions on 30 September 2026, with the global rollout continuing into 2027. Organisations need a D-U-N-S number and a verified website, so it's worth starting early.
On money, the flat 30% era is over on Play as well. From mid-2026 the standard fee on one-off in-app purchases dropped to 20%, subscriptions are 10%, the first $1 million of annual earnings is charged at 10%, and a separate billing fee applies where Google handles payments. For most UK developers the real take rate is well below the old headline. We cover the full picture in our App Store fees piece.
An Android app is not a one-off build. Google runs an annual OS cycle, the target-API requirement rolls forward every year, and there's a steady stream of SDK and policy updates to keep up with. An app that stops being maintained drops down the rankings, loses discoverability to new users, and eventually can't be updated at all. We offer app support and maintenance as a continuing relationship precisely because this is ongoing work, not a finishing line. Budget for it from day one.
Creating an Android app in 2026 means: validate the idea cheaply, design for Material 3 Expressive from the start, decide early whether you also want iOS (in the UK you probably do), build with on-device AI where the hardware allows, clear the API 36 target deadline and the verification and testing gates so you're not blocked, and commit to maintaining it. None of it is hard with the right team, and most of the failure modes are avoidable.
If you'd like help turning an idea into a real Android app, or working out whether to go native or cross-platform, get in touch. We'll give you a straight answer.
How much does it cost to create an Android app?
It depends on complexity, but the Google Play Console is a one-time $25 fee, and the build itself ranges widely. We cover real UK figures in our guide to how much it costs to build an app. The biggest cost lever is scope, so validating what to build first usually saves the most.
Do I need to use Kotlin to build an Android app?
For a fully native Android app, yes, Kotlin with Jetpack Compose in Android Studio. But you can also build for Android (and iOS at the same time) with a cross-platform toolkit from a single codebase, which is often more cost-effective if you want both stores. Native is the right call when you need the exact Android 16 look and the newest platform APIs.
What is the 12-tester rule on Google Play?
New personal Play Console accounts must run a closed test with at least 12 testers, opted in for 14 continuous days, before applying for production access. It applies only to personal accounts created after late 2023; organisation and older accounts are exempt. It's a real planning consideration for first-time solo publishers.
Why do Android apps get rejected or blocked from updating?
The most common avoidable reasons in 2026 are failing to target the required API level (API 36 from 31 August 2026), missing developer identity verification, policy issues around data and permissions, and not completing the closed-testing requirement on new accounts. Almost all of it is avoidable with proper checks before you submit.