In aircon cleaning, house cleaning, moving or repair, many chats open with "how much?". You can't answer in one line, because the price depends on details the customer hasn't given yet, so the chat becomes a slow back-and-forth.
A structured quote flow fixes that order. The customer answers the same questions in the same sequence, written rules turn the answers into a price range, and anything unusual goes to a person. Most of this is business work, not code. The last section is honest about where code starts.
Step 1: Read your own "how much?" chats
Scroll back through recent price questions. For each one, note:
- Which questions you had to ask back
- Which answer actually changed the price
- Whether you ended up saying "I need to see it first"
You're looking for price drivers: the few facts that move the number. For aircon cleaning: unit type, quantity, outdoor units, stairs. For house cleaning: floor area, bathrooms, move-out condition.
Details that never changed the price (preferred day, parking) can wait until after the range.
Step 2: Turn the drivers into a question order
Put the questions in a fixed sequence:
- Disqualifiers first. Ask the service area before anything else.
- Biggest price driver next, then quantity, then add-ons, then access (floor, elevator, height).
- One question per message.
- Offer fixed choices, such as "Split / Window / Not sure". They're easier to price than free text.
- Always include an escape option like "Not sure" or "Other", and treat it as a hand-off.
- Stop when the price is decided. If a question doesn't change the range or the hand-off decision, move it after the quote.
Step 3: Write the price rules as a table
Most owners price in their heads. Write that logic down so anyone gives the same answer. A spreadsheet with a few small tables works well:
| Table | What goes in it |
|---|---|
| Price list | Base price per unit type, plus a "needs a person" flag for types you never quote remotely |
| Add-ons and discounts | Per-unit add-on fees, a multi-unit discount threshold, a stairs surcharge, and how wide the range should be (for example ±10%) |
| Hand-off rules | Areas you serve, and words that always go to a person |
| Test sheet | Old enquiries typed back in, with the expected result next to them |
A worked example with made-up numbers: 3 split units at 1,800 each is 5,400. Outdoor units at 600 each add 1,800. A 3-unit discount of 200 each removes 600. Fifth floor with no elevator adds 300. The total is 6,900, and a ±10% range gives roughly 6,210–7,590.
A range leaves room for what only shows up on site. Always pair it with "final price confirmed on site".
Step 4: Decide hand-off before you decide prices
Hand-off rules stop you quoting jobs you shouldn't. Common triggers:
- Outside your service area
- Unit or job types you always inspect first
- Safety or access issues, such as unusual installation height
- Messages mentioning discounts, warranties, leaks, damage or complaints
- The customer picked "Not sure" or typed something that doesn't fit
The hand-off message should say what happens next without promising a price, for example: "This one needs a technician to check. We've passed it to our team and will reply during opening hours." Only state a reply time you can actually keep.
Step 5: Write the range reply once
Draft the message the customer sees at the end, and reuse it:
Based on your answers (3 split units, outdoor units included, 5th floor, no elevator), the estimate is about 6,210–7,590. Final price is confirmed on site. Reply "book" to choose a time, or "person" to talk to our team.
Keep it factual: no urgency, no discounts you haven't decided on.
Step 6: Run the flow with built-in features first
On a LINE Official Account, the account manager has a greeting message, keyword-triggered auto-responses and a rich menu. A "Get a quote" menu button can send a keyword, and the auto-response can reply with your first question and the list of choices. Staff follow the written order and calculate with the spreadsheet.
On the WhatsApp Business app, a greeting message, saved quick replies (for example a /quote shortcut holding the question list) and chat labels such as "Quote pending" cover the same pattern by hand.
It isn't automation, but it's consistent: same questions, same table, clear hand-offs.
Step 7: Test on old chats before changing anything
Run ten past enquiries through your sheet. Where the result differs from what you charged, fix the rule or add a hand-off trigger. Repeat whenever prices change.
Where code starts (be honest with yourself here)
Everything above is owner work. Having a bot ask and calculate by itself needs a webhook: a small web service that receives each message and replies.
- LINE: the Messaging API sends messages to a webhook you host. Setting that up means creating a channel, storing secrets and deploying code. That needs basic terminal and Node.js comfort, or someone technical to help for an afternoon.
- WhatsApp: automated flows go through the WhatsApp Business Platform (Cloud API) or a third-party tool built on it. The method in this article still applies, but the tooling and setup differ.
Pricing doesn't need AI. A plain rules engine is predictable, testable and can't invent a price.
Templates if you want a starting point
These are AI-assisted templates from one itch.io store. They are starting points to review and adapt, not finished systems or guarantees of results. The sample shop and prices in them are fictional.
- Free AI Chatbot Client Intake Pack: a questionnaire, knowledge-base CSV and go-live checklist. https://harrisonlineai.itch.io/ai-chatbot-intake-pack
-
LINE AI Quote Helper Lite (free, MIT): a rules-based quote engine, a
rules.jsonfile, logic tests and a static web demo of a LINE-style chat. You can open the demo to see the question flow; it doesn't connect to a real LINE account. https://harrisonlineai.itch.io/line-ai-quote-helper-lite - LINE AI Quote Helper Kit (US$24): adds an owner-facing Excel rules sheet with a test calculator, a Cloudflare Worker webhook for the LINE Messaging API, a client FAQ template and a bilingual setup guide. It is LINE-only and does not call any AI model; quoting is deterministic rules. Connecting the webhook needs basic terminal and Node.js skills, and you bring your own LINE credentials. https://harrisonlineai.itch.io/line-ai-quote-kit
Neither kit supports WhatsApp.
This article was written with AI assistance.
Top comments (0)