The app features that take the longest to build are rarely the ones you see in a demo. They're the ones where data changes across several phones and a server at once, or where a mistake costs someone money: real-time chat and tracking, payments and refunds, offline mode with sync, and the admin panel and backend behind it all. A login screen and a chat screen can look equally simple in a Figma prototype. One of them is a form. The other needs a server holding open connections, message ordering, delivery receipts, push notifications, media handling and a way to block abusive users.
What moves a quote is the number of things that can go wrong, not the number of screens. Each of these features has a "happy path" that is quick to build and a long tail of failure cases: the payment that's neither successful nor failed, the driver whose phone kills the location service, two people editing the same record offline. Those cases are where the hours go. If you want to see how that turns into figures, our guide on mobile app development cost in India breaks it down. This article explains why.
Real-time features: chat, live tracking and notifications
Chat is the classic underestimate. Founders picture a list of bubbles. The build needs:
- A connection layer, usually WebSockets or a service like Firebase Firestore, that keeps messages flowing while the app is open
- Message storage, ordering and pagination so old conversations load without freezing the screen
- Sent, delivered and read states, plus typing indicators if you want them
- Image, voice note or file uploads, with compression so you're not paying to store full-size phone photos
- Push notifications for when the app is closed, and correct badge counts
- Reporting and blocking. Apple's App Store guideline 1.2 requires apps with user-generated content to offer filtering, reporting and blocking, so this isn't optional if users can message each other
Group chat multiplies all of this. Every rule about who can see what has to be enforced on the server, not just hidden in the app.
Live tracking, for delivery, cabs or field staff, runs into the phone itself. Background location needs extra permissions on both platforms, and Google Play asks for a declaration explaining why your app needs it. Many Android phones sold in India, especially Xiaomi, Oppo and Vivo models, have aggressive battery optimisers that stop background services without warning. You end up writing workarounds, showing users how to whitelist the app, and testing on real devices rather than emulators. Then there's GPS jitter to smooth out, battery drain to manage, and Google Maps Platform usage that is billed per request, so careless polling shows up on your monthly bill.
Notifications sound like a checkbox. In practice you're handling FCM and APNs tokens that expire, deep links that open the right screen, user preferences, and the difference between transactional and marketing messages. If you want SMS as a fallback in India, you also need DLT registration of your sender ID and templates under TRAI rules, which involves paperwork as well as code.
Payments, wallets and refunds
Adding a Razorpay, Cashfree or PayU SDK is the easy part. The time goes into making sure your records always match the gateway's.
- Server-side verification. The app can't be trusted to say "payment succeeded". Your backend has to verify the gateway's signature and listen for webhooks.
- Pending states. UPI payments sometimes sit in limbo. The user's bank shows a debit, your app shows nothing. You need a pending state, retry checks and clear messaging, or your support inbox fills up.
- Refunds. Full, partial, cancelled orders, failed deliveries, refunds to the original method versus store credit. Each needs rules, an approval flow and correct accounting.
- Invoices and GST. If you sell to businesses, they'll want proper tax invoices generated automatically.
- Reconciliation. Someone in your team will need a report that matches orders to settlements.
Wallets are a bigger step than most founders expect. A closed-loop wallet, usable only for your own goods and services, doesn't need RBI authorisation, but it still needs a proper ledger where every credit and debit is recorded and balances can never go wrong. Anything that lets users pay third parties or withdraw cash moves into prepaid payment instrument territory, which needs RBI authorisation. Also note that Apple requires its own in-app purchase system for digital content and subscriptions, while physical goods and services can use your own gateway. Picking the wrong model means rework.
Offline mode and data sync
"It should work without internet" is one sentence in a brief and one of the hardest things to build well. There are two very different versions:
- Read-only offline. The app caches recent data so users can view it with no signal. This is manageable.
- Full offline editing. Users create and change records offline, and everything syncs later. This is where projects stretch.
Full offline needs a local database on the phone (SQLite, Room, Core Data, or something like WatermelonDB or Realm), a queue of pending changes, IDs generated on the device so records don't clash, and rules for conflicts. If a field agent edits a customer's address offline and the office edits it online, which version wins? Someone has to decide, and the code has to apply that decision every time. Add database migrations when you ship updates, partial syncs that fail halfway, and testing on flaky networks, and you can see why this is rarely a small line item.
The admin panel and backend nobody sees in the demo
The customer app is what you show investors. The admin panel is what your team uses every day, and it can be as much work as the app itself. A typical one needs:
- Roles and permissions, so support staff can't issue refunds and a city manager only sees their city
- Management screens for users, orders, products, content and banners
- Support tools: search a user, see their history, resend an OTP, cancel an order
- Reports and CSV exports your accountant will ask for
- Audit logs showing who changed what
Behind that sits the backend: API design, hosting, a staging environment, backups, crash and error monitoring with tools like Firebase Crashlytics and Sentry, and a force-update mechanism so you can retire old app versions. None of it appears in a demo. All of it decides whether the app survives its first busy month.
How to use this when you're scoping your app
You don't need to build every hard feature properly in version one. Some practical trade-offs:
- Chat: a hosted service like Sendbird or Stream saves build time in exchange for a monthly fee. If chat isn't your core product, that's usually the right call. If messages only need to update every few seconds, simple polling may be enough to start.
- Payments: use the gateway's hosted checkout and process refunds manually from the gateway dashboard until volume justifies automation. Skip the wallet unless it's central to how you make money.
- Offline: start with read-only caching. Build full offline editing only when real users in real conditions prove they need it.
- Admin: a framework admin such as Django Admin, or a tool like Retool or Strapi, can cover internal needs early. Custom-design it later.
When comparing quotes, ask each agency how they handle a payment webhook that never arrives, a phone that kills background location, and a sync conflict. Vague answers usually mean those cases aren't in the price, and they'll show up later as change requests.
If you'd like a rough estimate for your own feature list, try the pricing calculator.
















