In-app chat support for a small app: build, buy or link out?
Short answer: Most small apps should start by linking out: a "Contact us" screen that opens a pre-filled email or a messaging chat, plus a help centre for common questions. Move to a support SDK (Intercom and Zendesk are two examples) when conversations become frequent enough that you need history, push alerts for replies and the user's account details next to each message. Building your own chat only makes sense when chat is part of the product itself, not just a support channel.
The question isn't really about chat bubbles. It's about four practical things: can the user reach you without leaving the app confused, do they find out when you've replied, do you know who they are without asking, and who answers at 9pm on a Sunday?
The three options compared
| Link out (email, messaging app, help centre) | Support SDK (e.g. Intercom, Zendesk) | Build your own chat | |
|---|---|---|---|
| Set-up effort | Small: a screen and some links | Moderate: add the SDK, configure push and security | Large: backend, real-time messaging, admin inbox |
| User stays in the app | No, they switch apps | Yes | Yes |
| User hears about your reply | Via their email or messaging app | Push notification from the vendor, once configured | Only if you build push |
| You know who's asking | Only what you pre-fill or they type | Account details attached, if you identify users | Fully, it's your data |
| Conversation history | In your inbox, not in the app | In the app and the vendor's inbox | Whatever you build |
| Ongoing cost | Low | Vendor subscription (check their current pricing page) | Your development and hosting time |
| Who handles it day to day | Your normal inbox | Vendor's agent workspace | A tool you must also build |
Option 1: Link out, done properly
Linking out sounds basic, but done well it covers most small apps for a long time.
Pre-filled email. On Android, Google's common intents guide shows composing an email with ACTION_SENDTO and a mailto: scheme so only email apps respond, with extras for recipient, subject and body. Pre-fill the subject with something useful ("Support: Order #1234") and add the app version, device model and the user's account ID at the bottom of the body. That alone removes the first round of "which version are you on?" emails. iOS can do the same with a mailto link or the system mail composer.
A messaging app. Many UK customers would rather message than email. A "Chat on WhatsApp" button opening a chat with your business number is easy to add; check WhatsApp's own click-to-chat help for the link format and remember the conversation then lives on a phone, not in a shared inbox, unless you use a business messaging tool.
Help centre first. Put your five most common answers above the contact buttons. Password resets, refunds, delivery times. A good help screen reduces messages more than any chat widget.
Weak spots: the user leaves the app, replies arrive somewhere else, and you can't easily see their in-app context.
Option 2: A support SDK
Support platforms offer mobile SDKs that put a messenger inside your app. A few details from the vendors' own docs show what's involved:
Installation. Intercom's iOS installation guide offers Swift Package Manager or a manual xcframework; it notes that CocoaPods support is deprecated, with new versions stopping before the CocoaPods trunk goes read-only on 2 December 2026. Zendesk's iOS getting started guide lists a Zendesk Support or Suite plan and iOS 16 or later as prerequisites and gives an approximate download size for its framework (around 7.5 MB, varying by version), so check the size impact on your app.
Push for replies. This is the feature that makes in-app chat work. Without it, a user asks a question, closes the app and never sees your answer. Intercom's push notification guide has you create an APNs key with Apple, upload it with your Key ID, Bundle ID and Team ID, then register device tokens in the app. Budget time for this on both iOS and Android.
Knowing who's asking, securely. Identifying the signed-in user means support sees their account straight away. But it has to be done securely: Intercom's messenger security page recommends JWTs signed on your server so bad actors can't impersonate users or read their conversation history, and says that once enforced, requests without a valid token are rejected. Whichever vendor you use, insist this is set up before launch, not after.
AI first answers. Several support platforms now offer an AI agent that answers from your help content before a person steps in. It can cover out-of-hours questions, but it's only as good as your help articles, so write those first and check what the vendor charges for AI resolutions on its current pricing page.
Weak spots: subscription costs that scale with use, a dependency on the vendor's SDK updates, and app size.
Option 3: Build your own
Building chat means real-time messaging, storage, push notifications, attachments, an inbox for your staff, assignment, read states and moderation. That's a product, not a feature. It's worth it when conversation is the service: a tutoring app where learners message tutors, a marketplace where buyers talk to sellers, a clinic app where patients message the practice. For "I have a question about my order", it's almost never worth it.
Don't forget the people side
Whatever you choose, decide before launch:
- Hours. Show them on the contact screen ("We reply 9am to 5pm, Monday to Friday"). An instant-looking chat that takes two days to answer is worse than an honest email form.
- Response target. Set one you can keep, and show expected reply times if the tool supports it.
- Who answers. One named person with a backup beats "the team".
- Where context lives. App version, device and account ID should arrive with every message, whichever channel.
Verdict
- New app, low volume, one person answering: link out. Pre-filled email plus a short help screen, maybe a messaging button.
- Growing app, daily questions, account-specific issues: a support SDK with push and secure user identity.
- Chat is the product: build it, and plan for it as a core feature with its own budget.
- In all cases: help content first, honest hours, and the user's context attached automatically.
FAQ
Will an in-app chat SDK slow my app down?
It adds code and size; Zendesk, for instance, publishes an approximate framework size. Load the messenger only when users open support, and test start-up time before and after.
Can I start with email and switch to an SDK later?
Yes. Keep the contact entry point in one place in the app so you can swap what it opens without redesigning screens.
Do users need to be logged in to chat?
Not necessarily, but identified users get faster help. If you identify them, use the vendor's secure method so nobody can pretend to be another user.
Does in-app chat replace a help centre?
No. A help centre answers the common questions instantly and feeds any AI first answers; chat handles the rest.
When you're weighing up support options for your app, we can map out the simplest set-up that fits your users.
Sources
- https://developers.intercom.com/installing-intercom/ios/installation
- https://developers.intercom.com/installing-intercom/ios/push-notifications
- https://developers.intercom.com/installing-intercom/ios/secure-your-messenger
- https://developer.zendesk.com/documentation/zendesk-web-widget-sdks/sdks/ios/getting_started/
- https://developer.android.com/guide/components/intents-common