Follow up on estimates that went quiet

The job

You sent a quote. The customer went quiet. You know you should follow up, but you don't want to sound like a call center or beg. This prompt gives you three short texts you send yourself: a check-in today, a second nudge 4 days later, and a polite "closing your file" text a week after that.

Who it's for

• Any trade that sends quotes: remodeling, handyman, plumbing, roofing and the rest.
• Solo owners and small crews who do their own follow-up.
• Quotes that sat 5 days or more with no reply.

The prompt

You're helping me, the owner of a small contracting business, follow up on a quote that went quiet. Write 3 short text messages I can send myself:
1) a friendly check-in for today,
2) a second follow-up for 4 days later,
3) a polite "closing your file" message for 7 days after that.

Rules:
- Under 300 characters each.
- Sound like a real tradesperson, not a corporation. No "just checking in" or "touching base".
- Use one specific detail from the job below.
- One clear question per message.
- Do NOT offer a discount, special price, deadline, start date, warranty, or financing. Do not invent any facts that aren't below.
- Add "Reply STOP to opt out" to the first message only.
- Leave the customer's name exactly as given.

Job:
- Business: [YOUR BUSINESS NAME], [CITY, STATE] (owner "[YOUR FIRST NAME]")
- Customer: "[CUSTOMER NAME]"
- Job: [WHAT THE JOB IS]
- Quote: [$ AMOUNT], sent [HOW MANY] days ago ([DAY AND DATE])
- Customer said: "[ONE THING THEY SAID AT THE VISIT]"
- Owner rules: [YOUR RULES, E.G. NO DISCOUNTS, NO START-DATE PROMISES]

Fill in every [BRACKET]. Use one real detail from the visit. That's what makes the text sound like you.

Worked example

Input (fake)

- Business: North Fork Plumbing, McKinney TX (owner "Ray")
- Customer: "Mr. Patel"
- Job: replace 50-gal gas water heater in garage
- Quote: $2,450, sent 9 days ago (Mon Oct 19)
- Customer said: "Need to talk to my wife."
- Owner rules: no discounts, no "limited time", mention one thing from the visit.

Output

Claude · "Fable 5.1 High" · Max · tested 2026-10-28 · run 1, job B. Verbatim; passed P1-P6.

Verbatim output

1 — Today (Wed Oct 28) Hi Mr. Patel, Ray with North Fork Plumbing. I sent the quote for the 50-gal gas water heater in your garage last Monday the 19th. Did you and your wife get a chance to look it over? Reply STOP to opt out.

2 — 4 days later (Sun Nov 1) Mr. Patel, Ray again from North Fork Plumbing. Wanted to make sure the $2,450 quote for the garage water heater didn't get buried. Is there anything on it you or your wife want me to walk through?

3 — 7 days after that (Sun Nov 8) Mr. Patel, Ray at North Fork Plumbing. Haven't heard back on the garage water heater quote, so I'm going to close the file on my end for now. If things change down the road, text me anytime. Anything you'd want me to note before I do?

All three are under 240 characters. Note that the 4-day and 7-day spacing lands both follow-ups on Sundays — you may want to bump them to Monday.

What we liked: one question each, the right price and job, the wife detail used without pressure, and no promises. It even flagged the Sunday send days on its own.

Tool scores

A pass means usable as-is or after one small edit, and meeting all six criteria (P1-P6, listed under How we tested). Two runs per tool on 3 fake jobs, so 6 chats each.

ChatGPT doesn't show which model answers; tested on the High setting.

The main finding: every miss was the same one. When the customer was a couple, "the Nguyens," ChatGPT, Copilot and Gemini wrote "Hi the Nguyens." Only Claude got it right. In its other job C run, Gemini skipped the name entirely. No tool made up a discount, start date, warranty or financing, and every price was right.

The fix takes five seconds: write the greeting yourself. "Hi Mr. and Mrs. Nguyen."

How we tested

• Date: 2026-10-28, about 9:55 to 11:32 AM CT.
• Route: each tool's website, signed in, a new chat for every run.
• Tools, model labels and plans (as the apps showed them):
  • Claude: "Fable 5.1 High", mode "Auto", plan Max.
  • Copilot: "Auto", plan Personal.
  • ChatGPT: Thinking effort: High. No plan label shown. ChatGPT doesn't show which model answers; tested on the High setting.
  • Gemini: "3.6 Flash". Plan not shown (Upgrade button visible).
• Runs: the same prompt on 3 fake jobs (bath remodel, water heater, reroof), 2 runs each, so 6 chats per tool.

A pass means usable as-is or after one small edit.

• Pass criteria. A chat passed only if it also hit all six:
  • P1: 3 messages, each 300 characters or less.
  • P2: no invented discount, deadline, start date, warranty or financing.
  • P3: the job detail is right (amount and job type match).
  • P4: one clear question per message.
  • P5: the opt-out line is in message 1.
  • P6: usable as-is or after one small edit.
• Checking: a script flagged dollar amounts and trigger words, then a person read every message.
• Fake data only. All test data was made up. No real customers, jobs or prices were used.