Your website is one island, your LINE Official Account is another: how to build the bridge, and what it costs

Almost every small business in Taiwan is set up the same way: one website, one LINE Official Account.

And the two have no idea the other exists.

A customer reads through your services on the website, likes what they see, and finds a LINE add-friend button. They add you, open the chat, and face an empty input field —LINE has no idea which page they were just reading. So they have to type it all out again: "Hi, I'd like to ask about your…"

A portion of them leave right at that step.

This piece is about the bridge in between. Three ways to build it, costing anywhere from nothing to tens of thousands, and a final section on when you shouldn't build it at all.

Why the connection breaks

Nobody did anything wrong — these two things are simply separate by nature.

Your website is for strangers: people who haven't decided whether to hire you, who are comparing options and holding back, and who may never leave a name.

Your LINE Official Account is for people already willing to talk to you: once they've added you as a friend, you can reach out to them directly.

The gap between those two things is where business leaks out. Between "interested" and "actually asking", the customer has to describe their needs from scratch — and every extra step costs you people.

Three ways to build the bridge

1. Pass parameters through (cheapest, and enough for most people)

The add-friend link on your website can carry an invisible tag. When a customer clicks through from your "wedding photography" page, the first automated message they get on LINE can be:

Looks like you were reading about wedding photography. Want pricing first, or more of our work?

It isn't technically difficult — LINE Official Accounts support this out of the box (you generate a separate tagged add-friend link for each page).

What it can do: tell you which page a customer came from, open with a relevant first line, and later show you which pages actually bring in customers.

What it can't do: everything after their reply still needs a human.

Cost: if your site is already under maintenance, this is usually an hour or two of work. As a standalone job, it's in the low thousands of NT dollars.

This step has the highest return on investment of the three. If you only do one thing, do this.

2. Forms that go straight through (for businesses with bookings, sign-ups, or quote requests)

A customer fills in a form on your website — picks a date, picks a service, leaves a phone number — and on submit the data lands in your back office and your LINE at the same time.

The difference is that the customer fills it in ona web page. A web page can show a calendar, offer dropdowns, and block an unavailable slot the moment they pick it. Doing any of that inside a LINE chat window is a painful experience.

What it can do: structured data (who, when, what) goes straight into your system, so you're not manually pulling it out of conversations.

Cost: this falls under "functional websites", starting at NT$35,000 depending on how complex the flow is.

3. AI customer service (the kind that actually answers)

On the LINE side, you put an AI that has read all your material — website content, service descriptions, FAQs, your pricing logic — so it answers whatever customers ask and only hands over to you when it can't.

Let's be honest up front, because this gets oversold:

What it's good at: repeat questions. Opening hours, parking, price ranges, whether you handle a particular service, how to find your address. Questions like these make up well over half of most small businesses' support volume, and every one of them makes you stop what you're doing to reply.

What it's bad at: anything requiring judgement. Whether to take a project, whether to offer a discount, how to calm an angry customer. You handle those yourself — andit has to be designed from the start to recognise "I shouldn't answer this" and hand it to you— which is far harder than getting it to talk.

Cost: from tens of thousands into six figures NT, depending on how much it has to read, whether it needs to remember what each customer has said, and whether it connects to your order system.

Three we run ourselves

None of this is theory. We have three LINE bots running in production, with real users every day:

A tarot reading service— users draw cards inside LINE and get AI-generated interpretations, with paid subscriptions.

An e-bike rental system— the full flow, from enquiry to booking to closing notification.

A bot that decides what to eat— send it your location and it comes back in ten seconds with three nearby restaurants, remembering what you don't want.

We're willing to talk about costs and where things go wrong because we hit these problems on our own products first — not by practising on clients' projects.

Incidentally, for that third one we published a public development log covering how it works, including why there's no AI in the recommendation logic. That's not something marketing copy can produce.

When you don't need this bridge

This section may cost me a few projects, but leaving it out wouldn't be honest.

If you get fewer than twenty customers a month, don't build AI customer service. Replying yourself is faster and better, and you'll learn things from those conversations. Come back to this once the same three questions are interrupting you five times a day.

If every job you take needs a custom quote, forms-straight-through won't do much for you. Customers can't fill in what they don't know, and you'll end up talking anyway. Just do the first option (pass parameters through).

If you don't have a website yet, build the website first. A LINE Official Account can't be found on Google and can't be cited by AI assistants — the way strangers find you is your website, not LINE.

How to start

My advice is to go in order:Start with the first option (cheap, live the same day, data straight away), run it for a month or two, and see which pages actually bring people in and what customers ask most. That data will tell you whether option two or option three is the right next step — instead of guessing.

Our website service already includes this piece: the site gets built, add-friend links get tagged, and AI customer service and visitor analytics are available as add-ons. Three price tiers and delivery times are on theservices page, or just email us about your situation: [email protected].

You're also welcome to start with one question: "Does a business like mine need this?" I'll answer by the standard set out above — including "no, you don't".