DEV Community

Harrison Hu
Harrison Hu

Posted on AI-assisted

How to Sell Fixed-Scope AI Chatbot Setup Packages to Small Shops (Without Selling Hours)

Small shops such as salons, cleaning companies and repair services often want "a chatbot" but rarely want to buy your hours. An hourly quote asks the owner to guess how long the work will take. A fixed-scope package fixes the outcome and the price, and shows where the work stops.

Here is how to build those packages step by step. All prices below are placeholders; set them for your own market.

1. Define the scope boundary first

Before you write any package, decide what one "unit" of chatbot work is. For small shops, a workable boundary usually looks like this:

  • One channel: a website widget, LINE, WhatsApp, Messenger or Telegram. Pick one per package.
  • One knowledge base with a capped number of question–answer pairs.
  • One hand-off path: when the bot is unsure, it passes the conversation to a named person and notifies them.
  • One training session and a short owner guide.

Write the boundary as nouns with numbers attached. Vague phrases like "full chatbot setup" are where unpaid hours come from.

2. Run an intake checklist before you quote

Don't quote from the first message. Send an intake questionnaire and quote after it comes back. Collect at least:

  • What the business sells, where it operates and its opening hours
  • The 10 questions customers ask most often, copied from real messages if possible
  • Questions the bot must never answer on its own (refunds, complaints, medical or legal topics)
  • How the bot should reply to "how much?": with a public price list, "from" prices, or a hand-off
  • Tone and languages
  • Who takes over when the bot can't help, and during which hours
  • Which channel accounts exist and who owns them

Treat the answers as your complexity signal. If the owner can't list their top questions, or prices change daily, the job is riskier than it looks. Sometimes the honest recommendation is the platform's built-in auto-reply feature.

3. Build two or three tiers, not ten

Keep the menu small enough that an owner can choose in one read. One common structure:

Package What it covers Placeholder price
A: FAQ assistant 1 channel, up to N Q&A pairs, human hand-off with notification, 1 training session $X–$Y, set for your market
B: Booking / quote intake A guided flow that collects job details (or a booking form), sends them to a spreadsheet and notifies the owner $X–$Y, set for your market
C: A + B combined Both, sharing one knowledge base and one notification flow Less than A + B bought separately

Publish "from" prices if you like, but give the exact quote after intake.

4. Write down what is out of scope

Out-of-scope items prevent more arguments than in-scope items do. Put them in every proposal, for example:

  • Payment collection, POS or ERP integration
  • More than one channel
  • Outbound broadcasts or marketing campaigns
  • Calendar sync across multiple calendars
  • Writing the business's policies or prices (the client supplies all content, and you don't invent answers)
  • Medical, legal or financial advice

Also state that third-party costs (messaging-platform fees, AI API usage, domains) are paid from the client's own accounts. Don't front them.

5. Handle change requests and add-ons explicitly

Scope will be tested. Decide your rules in advance:

  1. The price covers exactly what is listed in the proposal.
  2. Anything new, such as another channel, another integration or more Q&A pairs than the cap, is quoted separately before work starts.
  3. Content revisions during the build are capped at a stated number of rounds.

Turn common requests into priced add-ons: "extra channel", "extra block of N Q&A pairs", "extra language", "after-hours reply flow". That turns "can you just…" into a yes with a price.

6. Agree on hand-over and acceptance criteria up front

Fixed scope only works if "done" is defined. Attach an acceptance checklist the client ticks off, such as:

  • [ ] The bot answers the top 20 questions correctly in testing
  • [ ] Unknown questions hand off to a human, and the owner receives the notification
  • [ ] Hours, prices and address match the intake sheet
  • [ ] The bot says it is an automated assistant when asked
  • [ ] The owner can edit an answer using the owner guide
  • [ ] The training session is complete

Before go-live, replay real past customer messages and test on a phone. The client should own all admin accounts, with you on a removable staff role. Tie payment to milestones, such as a deposit to start and the balance on acceptance.

7. Offer a maintenance retainer as a separate product

Prices, hours and promotions will change after launch. Offer an optional monthly care plan:

  • Up to N knowledge-base edits per month
  • A basic uptime or function check
  • A short monthly summary
  • A clear cancellation notice period

Price it as $X–$Y per month, set for your market. A client who declines still gets a working hand-over and owner guide.

8. Pricing logic: what actually drives the number

You still need a way to reach a number. A simple worksheet:

  1. Estimate your honest delivery hours for each package.
  2. Multiply by your target day or hourly rate.
  3. Add a risk buffer for revisions and client delays.
  4. Round up to a published "from" price.

Then adjust for the factors that really change effort and risk:

  • Number of intents / Q&A pairs: more topics mean more content to write, review and test.
  • Number of channels: each channel adds its own setup, testing and quirks.
  • Languages: each language needs its own tone review and testing.
  • Hand-off complexity: one owner's phone is simple, while routing across a team of staff is not.
  • Quote logic: price ranges that depend on area, unit type or add-ons need rules and testing.

Finally, sanity-check against value: what does the owner hope to get back (fewer repeated replies, fewer missed after-hours inquiries)? You can't promise those outcomes, but they help both sides judge the price.

Resources

If you want templates for these steps, the following are AI-assisted templates from the harrisonlineai itch.io store. They are starting points to edit, not guarantees:


This article was written with AI assistance.

Top comments (0)