Blog · 2026-09-17

Building a LINE Bot: Costs, Process, and Build vs Outsource (2026)

When the console is enough, when a LINE bot is genuinely necessary, how long a self-build takes, what outsourcing costs, and four failure paths that will cost you money if you get them wrong.

The conclusion first: if the console can do it, use the console. Do not automate for the sake of automating. The situations that genuinely call for a LINE bot are rarer than most people assume.

But once you cross that line, a bot does things the console cannot do at all. This guide draws the line, then covers cost and implementation.

When the console is enough

The LINE Official Account console does all of this natively, with no code:

• Welcome messages for new friends, keyword auto-reply, business hours switching
• Rich Menu, including scheduled start and end
• Broadcasting and basic segmentation
• Coupons and loyalty cards
• One-on-one chat and live support

If everything you need is on that list, set the console up properly and stop. Details are in LINE Auto-Reply Setup and the complete LINE Rich Menu guide.

When a bot is genuinely necessary

Four situations where the console stops being enough.

There is also a money argument. Reply API messages through the Messaging API are on LINE's published list of free message types, so responding costs nothing against your allowance. Move a large volume of questions onto bot responses and your message bill falls.

What a LINE bot costs

Costs break into three parts, and judging by one of them alone will mislead you.

Self-building looks cheap on direct costs. A small cloud instance runs a few hundred a month. The real cost is time. Getting the integration working takes three to five days. Handling the failure paths, plus auto-start and backups, usually takes two to three weeks. After that, every LINE spec change, every host restart and every certificate expiry is yours to deal with.

Outsourcing has a one-off project fee that varies enormously with complexity. A simple keyword Q&A bot is one price. Connecting an order system, doing member binding and handling payments is another. Three things to nail down in the conversation: who pays for hosting, who maintains it after launch, and who owns the source code.

A subscription platform is the third route. No development, no servers to keep alive, a monthly fee, and flexibility limited to what the platform offers. If your needs are standard member management and support automation, this is the fastest path.

The process for a self-build

If you are building it yourself, the flow looks roughly like this:

1. Create a Provider and a Messaging API channel on LINE Developers, and take the channel access token and channel secret.
2. Stand up a public HTTPS endpoint as your webhook to receive events from LINE. A tunnelling tool works for local development, but production needs a fixed address and a certificate.
3. Verify the signature. Every webhook request from LINE carries a signature. Verify it, or anyone can forge events against your endpoint.
4. Handle the events. Adding a friend, sending a message, tapping a menu and blocking are all different event types. Handle them separately.
5. Reply. Responding through the Reply API is free, pushing through the Push API is billed. Keep the two clearly apart.

Step three is the one most often skipped and the most dangerous. An unverified signature means your bot is open to the world.

Four failure paths that will cost you money

The integration itself is not hard. These four are. They come from building membership bots ourselves.

One, binding must verify ownership. A user typing a member account into the chat box is not binding. Use a one-time verification code or admin approval. Otherwise anyone can bind someone else's account to their own LINE and walk off with their rewards.

Two, pick the right key for duplicate prevention, and let the database enforce it. To prevent the same member checking in twice in one day, the unique key is member plus date. And it belongs in a unique index in the database, not an if statement in your code. Double taps, network retries and two simultaneous requests can all slip past application logic. A unique index will not let them.

Three, anything that touches money has to be idempotent. Payouts, points issuance and refunds all need a unique reference number, with a repeated reference crediting only once. The dangerous case is not failure, it is uncertainty. When a request times out the money may or may not have landed. You cannot blindly retry those and you cannot mark them failed. Query the record, confirm its state, then decide.

Four, hard-code the order: confirm the credit landed, then send the push. A duplicate push costs you one extra message. A duplicate credit is a straight loss. The retry strategies for the two sides have to be separate.

The fuller system design, covering how to build the mapping table, how to pick unique keys for check-ins and prize draws, and the auto-start, alerting and backup trio for cloud deployment, is in Automating check-ins, payouts and prize draws.

Three things you need before launch

A bot is a persistent service receiving messages around the clock. Running it on a machine in the office is a single point of failure.

1. Auto-start and crash recovery. The service should come back up when the host reboots, and restart itself when the process dies.
2. External monitoring and alerts. A dead process cannot report that it is dead. Alerting cannot rely on the service itself, so use an external liveness check that notifies your phone the moment it goes down.
3. Automatic data backups. The binding table is the life of this system. Back it up on a schedule, somewhere other than the host.

Miss any one of these and it is not really launched.

Would rather not deal with any of this? Our platform handles it

We packaged all of the above into the LINE OA platform, so hosting, signatures, idempotency and backups are not your problem:

• API integration: connect the bot to your own order or membership system, included in the Professional plan
• 8-stage funnel tracking: every friend sorted into a stage automatically, with no state machine to write
• CRM contact management: user profiles, behaviour trails and tags in one place
• AI support agent: RAG knowledge base plus an LLM, handing off to a human when it cannot answer
• Keyword auto-reply: keyword, event and condition triggers
• Segmented push: pick recipients by stage and tag

Three tiers, monthly fee in USDT, no lock-in:

We also take on fully custom system builds. Specs and process are at LINE OA platform and custom automation development.

FAQ

Do I need to know how to code to build a LINE bot? It depends what you want. Keyword Q&A, Rich Menus and broadcasting are all in the console and need no code. Connecting your own systems, doing identity binding and triggering on events needs development work, or a ready-made subscription platform.

What does a LINE bot cost? Self-building means a few hundred a month for hosting, but development and maintenance time is the main cost, and handling the failure paths usually takes two to three weeks. Outsourcing varies enormously with complexity, and you should nail down who pays for hosting, who maintains it and who owns the code. Subscription platforms start at $29 USDT a month.

Do bot messages cost money? Responses through the Reply API are not billed, which is on LINE's published list of free types. Outbound pushes through the Push API are billed, counted per recipient exactly like a broadcast. Details are in the complete LINE broadcast guide.

Can a bot get my account suspended? Not if you follow the official API rules. What you should avoid are add-ons claiming to automate a personal account. Those violate the terms of service and carry a real risk of suspension.

The API behaviour and billing rules in this article come from the LINE Developers documentation and the LINE Official Account manual. Specs change, so treat the current information in the official docs as authoritative.