Follow-Up System for Small Business: Quotes, Rebooking, Reviews

You sent the quote and they said it looked good. Then the week happened. Nobody called back, you meant to check in, and by the time you did they had hired someone else. Same story with the client who missed Tuesday and never rebooked, the regular who stopped coming in, and the great job that never got a review. That is not a lead problem. It is a follow-up problem, and it is usually the first thing I fix.

Here is the follow-up system I set up for owners: a daily list built from tools you already have, a timing ladder for each type, AI drafts in your own voice, and one rule that never bends. Nothing sends until you say yes.

The four follow-ups that leak money

Most owners I talk to have the same four gaps. Each one is a person who already raised a hand, so a short, well-timed message is often all it takes to get an answer either way.

• Unanswered quotes. The estimate went out, the customer went quiet, and nobody asked twice. Silence is not a no. They may be comparing, waiting on a spouse, or busy.
• No-shows and late cancels. The slot is gone and the person feels awkward. Without a quick, friendly rebook message, they drift.
• Lapsed customers. The regular who came every few months has not been back. Nothing went wrong. Nobody reminded them.
• No review ask. The job went well and nobody asked. New customers read those reviews before they ever call.

Trades feel the first most, front desks the second, shops and restaurants the third and fourth. The system is the same for all of them.

Build the follow-up list from what you already have

You do not need new software. The list already exists in four places, and the job is to pull it into one sheet, one row per person.

• Your CRM or job software. Field service tools such as Jobber or Housecall Pro, or a general CRM, track quote status. Every quote sent but not won or lost goes on the list.
• Your booking or reservation tool. Past visits, cancellations and no-shows. Look for an export or a report with the last visit date.
• Your spreadsheet. If quotes live in a sheet, add two columns: date sent and last contact.
• Your inbox and texts. Search sent mail for quote, estimate and proposal. Anything from the last 60 days with no reply goes on the list.

1. 01

   Pull the four sources

   Export what each tool gives you; messy is fine. A booking export and your quote sheet are plenty for week one.

2. 02

   Make one row per person

   Name, best way to reach them, the language they write in, type (quote, no-show, lapsed, review), what it was for, last contact date, and notes a stranger should read before writing to them.

3. 03

   Record consent

   Note whether they agreed to texts and where: a booking checkbox or a line on the signed estimate. No consent on file means email or a call, not a text.

4. 04

   Mark do-not-contact rows

   Anyone who asked you to stop, anyone in an open complaint, anyone going through something hard. These rows never get a draft.

5. 05

   Sort by type and date

   Oldest quote first. The AI drafts from this list every morning, and it is the only list anyone works from.

The timing ladder for each type of follow-up

This is the ladder I start with. Adjust it to how your customers buy: a kitchen remodel decides slower than a drain cleaning. One rule never changes. The moment someone replies, books or says no, they come off the ladder.

The last quote touch does more work than it looks. Asking whether to close the file makes answering easy, and a no is useful too: the quote leaves your head, and you learn something about your price or timing.

How to draft follow-ups in your voice with AI

Automated follow-up sounds robotic because a template wrote it, not you. The fix is to show the AI how you already write. Collect 10 to 20 real messages you have sent customers, texts and emails. Strip phone numbers, addresses and anything private. Give them to the AI once, in a saved project in Claude or ChatGPT, so you are not pasting them every morning.

Examples beat descriptions. Telling an AI to sound friendly gets you a greeting card. Showing it ten of your texts gets you your length, your sign-off, and the way you actually talk about price.

You draft customer follow-up messages for my business, [business name]. I approve every message before it is sent. You never send anything.

Below are real messages I have written. Match them: length, greeting, sign-off, punctuation, how formal I am, and the words I use. Do not add phrases I never use. No exclamation marks unless my examples use them.

Rules:
1. Never invent a price, date, discount or promise. If a message needs one and it is not in the record, write NEED: price or NEED: date in its place and keep the rest of the draft short.
2. One reason to reply per message. One question at most.
3. Use one specific detail from the record (the job, the item, the visit) so it reads like I remember them.
4. If the customer wrote to us in Spanish, draft in Spanish, in the same voice.
5. If the notes mention a complaint, a loss, or anything sensitive, do not draft. Write FLAG and the reason.

My examples:
[paste 10 to 20 messages here]

Here is today's follow-up list. For each row, draft the message for the touch that is due, using the timing below. Return a table with: name, type, touch number, channel, draft, and any NEED or FLAG.

Timing:
Quote: day 2, day 5 to 7, day 14 (offer to close the file).
No-show: same day, next day, day 7.
Lapsed: when they are due back, then once two weeks later.
Review ask: day 0 or 1, then one reminder on day 3 to 5. Same wording for every customer.

Skip anyone marked do-not-contact, anyone who has replied since the last touch, and any text row where consent is blank.

[paste today's rows]

Quote, day 2 (text)
Hi [first name], it's [your name] with [business]. Checking that the quote for [job] came through. Happy to walk through any of it. Reply STOP to opt out.

Cotización, día 2 (mensaje de texto)
Hola [nombre], soy [su nombre] de [negocio]. Quería confirmar que le llegó la cotización para [trabajo]. Con gusto le explico cualquier detalle. Responda STOP para no recibir más mensajes.

No-show, same day (text)
Hi [first name], we missed you today. No problem at all. I have [time 1] or [time 2] open this week if either works for you.

Cita perdida, mismo día (mensaje de texto)
Hola [nombre], hoy no le vimos en su cita. No se preocupe. Tengo [hora 1] o [hora 2] disponibles esta semana, si alguna le sirve.

Quote, day 14 (email)
Subject: Your [job] quote
Hi [first name], I don't want to keep filling your inbox. Should I close this quote out, or keep it open for you? Either answer helps me plan the schedule.

The approval gate: nothing sends without your yes

This is the rule I do not bend, and it is why owners keep trusting the system after week one. Every draft waits for a person. The AI does the remembering and the writing. You do the deciding.

A draft can get a price wrong, miss that the customer called yesterday, or nudge someone whose family just had a loss. One wrong text in writing can undo years of goodwill. The gate also keeps you inside the consent rules, because a person sees every message first.

In practice it is a short morning routine:

• Drafts land in one place each morning: a sheet, a drafts folder, or one message to your phone.
• You approve, edit or skip each one. Skipped rows come back tomorrow.
• Approved messages go out from your own number and email, so replies come to you.
• A reply takes the person off the ladder and puts the conversation on your desk.

Text and email consent rules for follow-up

Check this part with your attorney before launch, because rules change and some states add their own. These are the federal basics as of October 2026, with the official sources at the bottom of the page.

• Texts need consent. Federal rules under the Telephone Consumer Protection Act cover automated texts much like automated calls. Marketing texts need prior express written consent, and reminders and other non-sales texts still need consent. The simplest habit is a checkbox at booking or on the estimate that says what you will text and how to stop.
• Stop means stop. Replies like STOP, QUIT, END, REVOKE, OPT OUT, CANCEL or UNSUBSCRIBE revoke consent, and so does any reply a reasonable person would read as a request to stop. The federal limit to honor it is 10 business days; build yours to stop at once. One short confirmation text is allowed, with no marketing in it.
• Mind the hour. Federal rules bar telephone solicitations before 8 a.m. or after 9 p.m. in the recipient's local time, and some states are stricter. I keep every send inside business hours.
• Register your texting number. In the U.S., businesses texting from software over a regular 10-digit number register it through their texting provider (A2P 10DLC). Unregistered traffic gets heavier filtering and extra carrier fees.
• Email has its own rules. Under CAN-SPAM, commercial email needs no prior consent, but it needs honest sender and subject lines, your physical postal address, and a working opt-out honored within 10 business days. Messages that only confirm a job the customer already agreed to are handled differently. I count quote follow-ups and win-backs as commercial.
• Health care front desks. Keep reminders to a name, a day, a time and how to reschedule. Leave the reason for the visit out of texts and voicemails, and check with whoever handles your privacy compliance before automating anything.

How to keep follow-up personal at scale

Volume is not the goal. The test for every message: would the customer believe you typed it yourself?

• One real detail from the record in every message: the job, the item, the date, the tech who came out.
• First names and short sentences, nothing that sounds like a campaign. Two or three sentences for a text, one short paragraph for an email.
• Sign as the person who will answer. If the front desk replies, the front desk signs.
• Match their language. If they wrote in Spanish, follow up in Spanish, and have a Spanish reader approve those drafts.
• No fake urgency: no made-up deadlines, no limited spots that are not real.
• Vary the touches. Day 14 should not repeat day 2.

How to ask for reviews without breaking the rules

Review asks are where well-meaning owners get into trouble. Here are the rules as the platforms and the Federal Trade Commission state them. Reread each policy before you launch and every few months after, because they change.

• Google. Its review policy prohibits incentives such as payment, discounts, or free goods or services. It also prohibits discouraging negative reviews, asking only the customers you expect to be happy, and asking staff to collect a set number of reviews. Reviews from current or former employees count as a conflict of interest.
• The Federal Trade Commission. Its rule on consumer reviews took effect on October 21, 2024. It bans fake reviews, bans incentives tied to a review saying something positive (or negative), and bans suppressing reviews with threats or intimidation. Courts can impose civil penalties for knowing violations. The Commission's staff also say that asking only happy customers could still break federal law, even though the rule does not name it.
• Yelp. Its content guidelines tell businesses not to ask customers for reviews at all. Leave Yelp out of your review asks and let those reviews come on their own.

In practice: ask every finished customer the same way, with the same link and nothing attached. If someone says something went wrong, fix it, and still do not steer them away from the review page. Reply to every review, good or bad, briefly and by name.

Before you turn it on

How to tell if your follow-up is working

Skip the vanity dashboards. Two numbers from your own records, counted weekly, tell you most of what you need.

• Reply rate, by type. Replies divided by follow-ups sent. A reply counts whether it is a yes, a no or a question, because every answer takes a person off your mind.
• Booked from follow-up. Jobs, visits or orders where the person booked after replying to a follow-up. Tag it at booking with one checkbox, so you are not guessing later.

Two more show whether the system is healthy. Opt-outs per hundred sends: if that climbs, you are sending too often or too generically. Drafts you rewrite: if you rewrite most of them, the AI needs better examples, not more rules.

Compare against your own baseline: the same numbers for the four weeks before you start, even roughly. I don't quote anyone a number before I have seen theirs.

What I'd build first

I'd start with quotes. They are people who already asked you for a price, the list is easy to pull, and the ladder is short. I connect the quote list from your CRM, job software or inbox, set up drafts in your voice from your own sent messages, and put one approval screen on your phone. Everything sends from your number and your email.

Once quotes run a few weeks without surprises, I add no-shows, then review asks, then lapsed customers, each with its own ladder behind the same gate. If missed calls are the bigger leak, start with answering every call and come back to this one.