A business needs a mobile app when being on a phone materially improves the task: using device capabilities, continuing without a connection, or supporting a repeated activity in a mobile context. If people only need to read information and send an occasional inquiry, a responsive website may be a better first investment.
The first release should complete one important journey well. Decide who it serves, where they use it, what happens when something fails, and who will operate it after launch. Platform and feature choices become easier when those answers are clear.
Give the user a reason to install
Describe the moment the app earns its place on a phone. A field worker may need to record a task away from a desk. A job seeker may want to build and export a CV. A navigation user needs orientation and guidance during a journey. “Our competitors have an app” does not tell a development team which experience matters.
Consider frequency, access to the device, and the cost of getting started. Installing an app, granting permissions, and creating an account are commitments. Explain the value before asking for them, and avoid requiring an account unless the product needs one. A prototype tested with the intended users can reveal whether the proposed journey solves their problem.
Choose platforms around the actual product
Native development uses a platform’s own tools and APIs. Cross-platform development shares part of the implementation across operating systems. Neither label settles the decision. Consider your audience’s devices, hardware needs, interface complexity, available team, and the ongoing work of testing and maintaining the product.
Sharing code does not remove platform differences. Permissions, purchases, background behavior, accessibility, and store releases still need attention on each target platform. Conversely, a single-platform first release can be reasonable if the intended audience is known and the scope is explicit. Do not describe an Android-only pilot as a completed iOS launch.
| Decision | What to establish | Why it matters |
|---|---|---|
| Audience | Which devices the first users actually have | Avoid excluding the intended pilot group |
| Device features | Location, camera, audio, or background needs | Prototype the hardest integration early |
| Operating model | Who maintains each supported platform | A release needs continuing ownership |
| Shared implementation | What can be shared and what remains platform-specific | Compare realistic total effort |
Swipe to compare columns, or focus the table and use the arrow keys.
Make the MVP a complete small journey
An MVP is useful when a person can reach a meaningful result and the business can learn from it. List the actions required for that result, then separate essential steps from helpful extras. For a document product, selecting a template, entering details, checking the result, and exporting it may form the core. An elaborate social feed would answer a different need.
One useful journey, four release decisions.
Business problem
Why use a phone?Name the person, context, and useful result.
Complete MVP
What must work end to end?Keep one journey, including recovery.
Device testing
What happens under pressure?Check permissions, interruptions, and connectivity.
First release
Who owns what comes next?Agree release access, support, and learning.
Pilot feedback can change the scope. A first release begins the next decision.
Do not remove recovery from the core journey. A failed upload, an accidentally closed screen, or an unavailable API should not silently discard valuable work. Define the saved state, the message shown, and the action the user can take. These decisions make a small release dependable without turning it into a large product.
Choose what you want to learn from the release and how you will learn it. A short interview about an unfinished task can be more informative than an undifferentiated download count. Any analytics plan should specify events, responsibility, and privacy implications before tracking is added.
Design for the conditions of use
Mobile use may involve bright light, one hand, interruptions, large text, or assistive technology. Put the primary action where it is easy to find and keep its meaning consistent. Explain the current state and the result of an action; color and animation alone should not carry essential information.
Review the main journey with a screen reader, enlarged text, and reduced motion. Check meaningful labels, reading order, focus, contrast, and touch targets. W3C’s accessibility guidance provides criteria for accessible content; native apps also need platform-specific accessibility review. A checklist supports testing but does not establish conformance by itself.
Creative products need particularly careful feedback. DrawBeat’s specified interaction maps horizontal position to time and vertical position to pitch. That relationship gives the interface a clear job: help a user understand what their marks mean. It is not evidence that every accessibility need has already been solved.
Reference: W3C: How to Meet WCAG 2.2
See in practice: DrawBeat: Sketch to Music
Treat integrations as part of the product
List each external service and what the app expects from it: identity, payments, content, maps, document processing, or business data. Establish who owns the account, how access is provisioned, and what happens when a response arrives late or not at all. The app’s design must include those states, rather than only a successful demonstration.
Separate the mobile interface from privileged business actions. A secret embedded in a distributed app is not a reliable place for privileged server credentials. Ask the team to explain which operations require a server, what the app stores locally, and how access can be revoked. Do not assume that adding a login screen answers those questions.
Define what offline actually means
“Works offline” can mean reading previously downloaded information, editing local data, or queuing an action for later. Specify each task separately. A document editor can retain local changes without promising that a map, a payment, or a newly requested remote record is available without connectivity.
If records can change on more than one device, decide how conflicts are resolved. Show when information was last synchronized and distinguish a saved draft from a submitted transaction. Android’s offline-first guidance describes local and network data responsibilities; the product still needs a business decision about what users can safely do while disconnected.
Test connectivity changing during a task, not just a phone that starts offline. Restore the connection after repeated retries, reopen the app, and check whether work appears once, remains pending, or needs the user’s attention. This makes offline behavior an acceptance criterion rather than a marketing phrase.
Test a release, not just a set of screens
Create a device and scenario matrix around the actual audience. Include supported operating-system versions, small screens, large text, slow networks, denied permissions, interruptions, and app restarts. Test the release configuration as well as development builds. Android’s core quality guidance is a useful starting reference, not a substitute for the product’s own critical journeys.
Arrange a pilot with representative users and a route for reporting problems. Give each issue a reproducible sequence, expected behavior, and enough non-sensitive context to investigate. Recheck a fix in the journey that originally failed; a change that repairs one screen can affect another.
Reference: Android Developers: Core app quality
Plan ownership before store submission
Agree who owns developer accounts, signing access, service subscriptions, and release materials. Prepare accurate descriptions, screenshots, privacy information, and any reviewer access needed for the real app. Store policies change; check the current requirements for each intended platform before submission. Approval and timing should not be promised in a development proposal.
A launch also needs a support path. Name the person who receives crash reports, the process for urgent fixes, and how compatibility updates are funded. Consider how users on an older app version interact with a newer server. A forced update, staged rollout, or rollback is a product decision with operational consequences.
Reference: Apple: App Review Guidelines
Ask for an estimate you can reason about
Cost is shaped by journeys, platforms, device capabilities, server requirements, integrations, offline complexity, testing, and release obligations. Artwork and content preparation, accessibility review, migration, support, and recurring services also need an owner. A screen count or a generic price band cannot express those differences reliably.
Bring the audience, one main journey, examples of the information involved, and known constraints. Ask what should be prototyped first, what can be deferred, and how the team will demonstrate completion. The proposal should make assumptions and responsibilities visible enough that you can compare it with another approach.
- Which user problem does the first release solve end to end?
- What happens if permissions, connectivity, or a provider fail?
- What is shared across platforms, and what needs separate work?
- Who owns accounts, source, release access, and ongoing support?
- How will acceptance and post-launch learning be assessed?
Further context
Sources & project evidence
Platform guidance changes. Consult the linked primary documentation when planning an implementation. Project details come from the linked Zyrox case studies; general recommendations are editorial guidance.
- W3C: How to Meet WCAG 2.2
Accessibility criteria supporting the reading, labeling, contrast, and interaction discussion.
- Android Developers: Build an offline-first app
Local and network data responsibilities; the examples here are planning advice, not a claim about every Zyrox app.
- Android Developers: Core app quality
Platform quality guidance to consult when assembling a product-specific test plan.
- Apple: App Review Guidelines
Current primary reference for an intended App Store submission; no iOS release of the cited Android projects is claimed.