Who this is for
Operators who said “I need a chatbot,” or who lose leads after hours — or on a quiet-but-ignored contact form. If you can't name the task yet, start at 01.
Name the job
About 5 min
A chatbot is allowed exactly one of these jobs. Pick yours before reading further:
- Read all five jobs below. Pick the one that matches tonight's pain.
- If you picked 5 (“look modern”), stop here — that's not a job, and the rest of this module will confirm it.
- 1. FAQ deflection
- 2. After-hours lead capture
- 3. Booking without back-and-forth
- 4. Answers from private data (inventory, schedule, job status)
- 5. “Look modern” — not a job. Stop here.
Book a chatbot demo before the self-test below. Demos sell; the checklist tells the truth.
Run the self-test
About 10 min
Check each statement that's true for your shop right now:
Bot-readiness self-test
- We get serious inquiries when nobody's working
- The answers exist in data we already store digitally
- Daytime form/SMS response is under ~10 minutes
- A named owner will read transcripts weekly
- We know what a qualified lead is worth
The rule: three or more unchecked → you are not ready for a public bot. Stay on the cheap stack below. That's not failure; that's a $200/month subscription you just didn't burn.
Run the 30-day stack
↓ Download the stack plan + self-test (PDF)
Try this
Do row 1 only this week: publish your top-10 FAQ where customers already look (website, Google Business Q&A).
You should see: fewer repeat calls on those topics within two weeks. If nothing changes, your problem isn't answers — it's response time. Fix that before row 2.
| Step | Action | Success signal |
|---|---|---|
| 1 | Publish top-10 FAQ + Google Business Q&A | Fewer repeat calls on the same topics |
| 2 | Instant text-back on missed calls | Callers stop rolling to the next contractor |
| 3 | Form auto-reply with a real next step | Not just “thanks for submitting” |
| 4 | Booking link for non-emergency work | Fewer “what days work?” threads |
| 5 | Weekly lead review (spreadsheet OK) | You know volume, source, response time |
Hold the day-30 review
About 20 min
Write one of three lines: keep the stack only / fix the response process / now consider live-data chat. If you consider chat, list the data source, the handoff, and the transcript owner before any vendor demo.
Success check
- Job named (1–4, not 5)
- Self-test scored
- At least steps 1–3 attempted, or consciously deferred with a reason
- Written day-30 decision (even “still in week 1 — revisit ___”)
Common failure modes
| Failure | Fix |
|---|---|
| Adding a channel while ignoring current leads | Fix response time before new capture |
| Bot demo before FAQ/SMS | Finish the stack first |
| No transcript/lead owner | Assign one before any AI channel |
| Treating booking as an AI problem | Publish real windows or a calendar link |
When to stop / don't buy yet
Skip chatbot SaaS if: little site traffic; you don't answer the leads you already get; the question is only “when can you come?”; you won't review transcripts. A clean form + fast text-back beats a clever bubble with a two-day human reply.
Adapted from the field note “AI chatbot for my business.”
That's the one chatbot worth building — inventory, schedule, or job status behind it. Bring the brief; we'll scope it fixed-price.