Best Prompts for Contractors and Home Services Businesses (2026): Copy-Ready Prompts for Leads, Estimates, and Reviews

The short answer: the best prompts for contractors are not clever tricks. They are short, specific instructions that turn the notes you already have from a call, a form, or a site visit into a finished message you can send in minutes. Start with the eight prompts on this page, one for each job that decides whether work gets booked: a new lead reply, a missed call text back, an estimate and proposal draft, a follow up on an unsold estimate, an appointment reminder, a review request, an invoice follow up, and a customer update during the job. Fill in the brackets, check every fact against your own notes before you send, and save the versions that work in a prompt library such as the tools in this directory's AI Prompt Tools category, so your office stops rewriting the same messages from scratch.

That speed matters because homeowners rarely wait. A speed to lead guide from RevCore Pro, summarizing Jobber's 2026 Home Service Trends Report based on a survey of 1,050 home service business owners in December 2025, found that only 20 percent of pros say they respond to leads within the hour, while 28 percent of customers expect an immediate response and another 28 percent expect a response within one hour. The same guide sets a practical target that this page follows: reply to a form, text, or chat lead within five minutes during business hours with a real next step, such as two appointment times, and text back a missed call within five minutes with a callback inside 15 minutes. A prompt cannot answer the phone for you, but it can remove the blank page that slows the reply down.

What makes a contractor prompt actually useful?

A useful contractor prompt does four things every time. First, it names the job in plain words, such as drafting a text to a homeowner who just requested a roof inspection, instead of asking for a generic marketing message. Second, it carries the facts the model is allowed to use, pasted from your notes, and it tells the model not to invent prices, dates, warranty terms, or materials that are not in those notes. Third, it sets the output shape: a length, a tone, and exactly what the message must include, such as two appointment times and one clear question. Fourth, it ends with a check step, asking the model to flag anything it had to guess so a person can fix it before sending.

Weak prompts fail in predictable ways. They ask for a professional email with no facts, so the model fills the gaps with invented details. They ask for five messages at once, so every message comes out thin. They leave out the trade, the town, and the customer's actual problem, so the output could belong to any business anywhere. Every prompt on this page is built to avoid those failures: one job per prompt, your facts in brackets, a fixed output shape, and a human check before anything reaches a customer.

If you want the wider tool picture behind these prompts, start with AI Tools for Contractors and Home Services (2026): The Full Stack, which covers the software categories a home services business uses day to day. This page is the companion to that stack: the words themselves, ready to paste into ChatGPT, Claude, Gemini, or the prompt tool you already use.

Which prompt should you use when a new lead comes in?

Use this prompt the moment a form, text, chat, or marketplace lead lands and you need a first reply that moves the job forward instead of just saying thanks. RevCore Pro's speed to lead guide makes the same point this prompt is built on: a reply with no next step barely counts, while a reply that confirms the job and offers two times gives the homeowner a reason to stop shopping. Paste your lead notes where the brackets are, and send the result within minutes, not at the end of the day. For the systems that deliver this reply automatically, see AI Tools for Lead Follow Up (2026).

You are helping the office at [business name], a [trade] company in [city and state]. A new lead just came in. Write a short first reply we can send by text and email.

Lead details from our form:
Name: [customer first name]
Job they described: [job in their words]
Property type: [single family home, rental, small commercial]
Urgency they stated: [emergency, this week, planning ahead, not stated]
Source: [website form, referral, marketplace, repeat customer]

Rules:
1. Confirm the exact job they described, in their words, in the first sentence.
2. Name the person sending it: [sender first name], from [business name].
3. Offer exactly two appointment times for a visit or call: [time option 1] and [time option 2]. Do not invent other times.
4. End with one easy question they can answer with a short reply.
5. Keep the text version under 60 words. Keep the email version under 120 words, with a subject line.
6. Do not invent prices, discounts, warranty terms, or arrival windows that are not in the notes above. If something is missing, leave a bracket for us to fill.
7. Tone: friendly, direct, local. No pressure and no hype.

After the two versions, list anything you had to guess, under the heading Check before sending.

The check list at the end is not decoration. On a busy day it is the difference between a fast reply and a fast mistake, because the model will show you the one detail it was unsure about instead of hiding it inside a polished sentence.

What should you send when you miss a call?

Send a short text from your business number within minutes, then call back inside the window you promised. RevCore Pro's guide puts it plainly: a missed call is a lead that already chose you, and many homeowners who reach a generic voicemail will not leave a message at all, they call the next company on the list. This prompt turns a name, a time, and a caller ID note into that text, plus a voicemail script for the callback if they do not pick up. Tools built for this exact job are compared in Best Missed-Call Text-Back Software for Service Businesses in 2026, and the receptionist options are covered in AI Receptionist Tools (2026).

You are helping [business name], a [trade] company in [city]. We missed a call and need to text the caller back right away.

Known facts:
Caller name if we have it: [name or unknown]
Time of the missed call: [time]
What we know about the job: [note from caller ID, voicemail, or form, or unknown]
Our callback promise: a person will call back by [time]
Person texting: [sender first name]

Write three things:
1. A text under 40 words that says who we are, that we saw the missed call, and that we will call back by the promised time. If the job is unknown, ask one question about what they need help with.
2. A second, shorter text we can use after hours that says when the office reopens: [opening time and day].
3. A voicemail script under 30 seconds for the callback, with the same facts and a clear way to reach us by text.

Rules: do not invent services, prices, or emergency coverage we did not state. If we do not offer emergency service, do not imply that we do. Keep the tone calm and human.

Run this prompt once, save the two text versions as templates, and you may never need to run it again for the same trade. That is the pattern for this whole page: use AI to build the template with care once, then let the saved template do the daily work at full speed.

How do you turn site visit notes into an estimate and proposal?

Start from qualification, not from a number. A 360 Media Consulting guide on turning estimates into paying customers advises contractors to qualify the job before pricing it, gathering the service needed, the property location, the desired timing, budget expectations where appropriate, whether all decision makers will be present, and whether the customer is collecting other quotes. The same guide warns against sending a bare number on larger jobs, because a vague quote forces the customer to compare price alone, while a clear proposal lets them compare scope, materials, workmanship, and risk. This prompt follows that order: it organizes your visit notes into a proposal with included work, excluded work, and open questions, and it refuses to price anything your notes do not price. Software that holds the full estimate workflow is reviewed in Best AI Estimate and Proposal Software for Contractors in 2026.

You are drafting a customer proposal for [business name], a [trade] company in [city and state]. Use only the site visit notes below. Do not add work, materials, prices, or warranty terms that are not in the notes.

Customer: [customer name], [property address or neighborhood]
Visit date and estimator: [date], [estimator name]
Problem or goal in the customer's words: [their words]
Scope we agreed to price: [included work, line by line]
Materials and products named on site: [materials, brands, or grades from notes]
Work we are excluding: [excluded work]
Price from our pricing sheet: [total or option totals, exactly as written]
Timing we discussed: [start window, duration estimate, or not discussed]
Warranty terms from our standard sheet: [terms or leave blank]

Write the proposal in this order:
1. A short opening that restates the customer's problem or goal in their words.
2. What is included, as a plain list matching the scope above.
3. What is not included, as a plain list, so there are no surprises later.
4. The price, exactly as given, with what the customer does to approve the work: [approval step, such as reply to this email or sign the attached copy].
5. What happens next after approval, in two or three steps.
6. A closing line naming [contact name] and [phone number] for questions.

Rules: if the notes are missing something a proposal needs, do not guess it. Write [NEEDS OFFICE INPUT: describe what is missing] in that spot instead. Keep the language plain enough for a homeowner who has never hired a [trade] company before.

The bracketed gap markers are the most valuable part of this prompt in daily use. An estimator coming off a roof can dictate rough notes, and the office gets back a clean proposal that shows exactly which two or three facts still need a human decision, instead of a smooth draft with a wrong number buried in paragraph three.

What should you say when an estimate goes quiet?

Follow up on a schedule, with something useful in each message, instead of asking again and again whether they saw your email. The 360 Media Consulting estimate guide recommends exactly that: replace memory with a follow up schedule matched to the job, and add context with every contact, such as a scope clarification, an open installation date, or a warranty detail. It also notes that higher value jobs can need several contacts over weeks, because financing, other decision makers, and timing all slow the decision down. This prompt writes that sequence in one pass, so the office can send the right message on the right day without drafting under pressure.

You are writing follow up messages for [business name] about an estimate that has gone quiet.

Facts:
Customer: [customer first name]
Job and scope in one line: [job summary]
Estimate sent on: [date], total: [total exactly as sent]
Their stated concern or priority on the visit: [budget, timing, disruption, comparing quotes, waiting on another decision maker, or not stated]
One useful detail we can add now: [open schedule date, warranty detail, scope answer, photo of a similar finished job we have permission to share, or phased option]
Sender: [sender first name and role]

Write four short messages, labeled Day 2, Day 5, Day 10, and Day 21:
1. Day 2: a check that they received the estimate, plus an offer to walk through it in a five minute call.
2. Day 5: lead with the useful detail above, tied to their stated concern.
3. Day 10: restate the scope in two lines, name the next open start date from our facts, and make approval one simple step.
4. Day 21: a polite closing message that keeps the door open and asks whether timing changed, with no pressure.

Rules: each message under 70 words, written for text or email. Do not offer a discount, change the price, or invent a deadline that is not in the facts. Do not repeat the same opening line twice.

Notice what this sequence never does: it never drops the price to buy a reply. The estimate guide cited above makes the same call, warning that routine discounting trains customers to wait you out, and that adjusting scope or phasing the work is usually the better answer when budget is the real objection. If the customer names budget as the blocker, ask the model for a phased version of the scope, not a cheaper version of the same scope.

How should you remind a customer about an appointment without sounding like a robot?

Remind them with the facts they actually need: who is coming, when, what to have ready, and how to reschedule without a phone call. A reminder that reads like a system notice gets ignored, and a missed appointment costs a crew a half day. This prompt writes the reminder in your voice, at three points that match how homeowners actually plan: when the job is booked, the day before, and the morning of the visit. If you want software to send these on its own, the booking side of the stack is covered inside the contractor tools guide linked earlier on this page.

You are writing appointment reminders for [business name], a [trade] company in [city].

Booking facts:
Customer: [customer first name]
Job: [job summary]
Appointment date and arrival window: [date], [window exactly as promised]
Who is coming: [technician or crew lead first name]
What the customer should have ready: [access, gate code step, pets secured, water shutoff location, cleared work area, or nothing special]
How to reschedule: [reply to this text, call this number]
Office phone: [phone number]

Write three reminders:
1. Booking confirmation, under 50 words: confirm the job, the date, and the arrival window.
2. Day before reminder, under 40 words: name who is coming, repeat the window, and list what to have ready in one short line.
3. Morning of reminder, under 30 words: we are on schedule, the window again, and the reschedule route if something changed overnight.

Rules: use the exact window from the facts every time, never a narrower one. Friendly and brief. No marketing, no upsell, no links that are not in the facts.

Save all three versions with your other templates. Crews change and seasons change, but the shape of a good reminder does not, so this is another prompt you run rarely and reuse daily.

When and how should you ask for a review?

Ask at the moment the work is done and the customer can see it, and make the ask effortless. A contractor reputation guide from Contractors Marketing Pros recommends asking every happy client at handover and again after the final walkthrough, sending one short text with the direct review link, keeping that text to two or three lines that thank them, name the project, and paste the link, and sending it from the person the client worked with rather than a generic inbox. The same guide is firm on the ethics, and this page is too: genuine reviews only, never bought, never traded for a discount, and never written by the model for the customer to post as their own. The prompt below writes your request, not their review. The tools that run this workflow are compared in AI Tools for Review Requests (2026).

You are writing a review request for [business name], a [trade] company in [city].

Facts:
Customer: [customer first name]
Project just finished: [project in a few words, such as the kitchen remodel or the new water heater]
Person asking: [first name of the person the customer worked with most]
Our direct review link: [paste the real link from our Google Business Profile]
Day we finished and handed over: [date]

Write two messages:
1. A text of two or three lines: thank them by name, name the project, say a Google review would mean a lot to our small team, and paste the link exactly as given.
2. One follow up message for a few days later, under 40 words, for someone who said they would leave a review and has not yet. Warm, brief, no guilt and no second reminder after this one.

Rules: do not write a suggested review for the customer to copy. Do not offer anything in exchange for a review. Do not claim a star rating, a review count, or a ranking for our business. Keep both messages in the voice of the person asking.

That first rule, never drafting the customer's words for them, protects you twice. It keeps you inside the review platform rules the guide cites, and it keeps your reviews sounding like your customers, which is why homeowners trust them in the first place.

How do you follow up on an unpaid invoice without damaging the relationship?

Send a clear, kind sequence that assumes an oversight before it assumes anything else, and states the facts the same way every time: the work, the amount from your own invoice, the date it was sent, and the one step that closes it. Most late payments in a small home services office are lost paperwork and busy weeks, not bad intent, so the tone should stay helpful while the facts stay exact. This prompt writes three steps, from a friendly nudge to a firm final notice your office can review before sending. The payment side of the stack is covered in AI Tools for Getting Paid Faster (2026).

You are writing invoice follow up messages for [business name], a [trade] company in [city].

Invoice facts, copied from the invoice itself:
Customer: [customer name]
Work completed: [job summary and completion date]
Invoice number: [number]
Amount due: [amount exactly as invoiced]
Invoice sent on: [date], due date: [date]
How to pay: [payment link, portal, or call the office at this number]
Who to contact with a question about the bill: [name and phone]

Write three messages, each under 80 words, for email, with a subject line:
1. Friendly reminder, a few days past due: assume it was missed, restate the amount and due date, give the payment step.
2. Second reminder, about two weeks past due: same facts, ask directly whether there is a question about the work or the bill we should resolve, and name the contact.
3. Final notice, about a month past due: same facts, state plainly that the balance is past due, give a specific date by which we need to hear from them: [date], and say the next step is a call from [owner or office manager name].

Rules: never change the amount, the dates, or the invoice number between messages. Do not add late fees, interest, or collection language that is not in the facts above. Polite and factual at every step.

Because the prompt locks the amount and dates to the invoice facts and forbids the model from adding fees, the sequence can be sent by whoever is covering the office that week without a fresh review of the numbers each time. Still read the final notice yourself before it goes out. Tone at that step is a business decision, not a writing task.

What should a customer update sound like while work is underway?

It should sound like news from the job, not like a status code. Homeowners get anxious when a project goes quiet, especially when the house is torn up and the crew has left for the day. A short update that says what got done, what happens next, and when you will be back answers the question they were about to text you. This prompt turns a crew lead's rough end of day notes into that update, in the same shape every time, so customers learn where to look for the one fact they need.

You are writing an end of day customer update for [business name], a [trade] company working at a customer's home.

Facts from the crew lead's notes:
Customer: [customer first name]
Project: [project summary]
Done today: [completed work, in the crew's words]
Found today that the customer should know: [issue, change, or nothing to report]
Happening next: [next step and when]
Crew returns: [date and arrival window, or office will confirm]
Anything the customer needs to do tonight: [run water, keep pets out of a room, leave a gate unlocked, or nothing]
Sender: [crew lead or office contact first name], reachable at [phone]

Write one update message under 90 words:
1. Open with what got done today, in plain homeowner language, translating trade terms from the notes into everyday words.
2. If something was found, state it plainly, say what it affects, and say what we recommend or what we are watching. Do not name a price for new work unless a price is in the notes.
3. Close with what happens next, when we return, and anything they need to do tonight.

Rules: do not promise a finish date that is not in the notes. Do not hide a problem the notes mention. Warm, short, and specific.

Of all the prompts on this page, this is the one customers mention later. It takes two minutes at the end of the day, and it is the message homeowners quote back when they explain to a neighbor why they would hire you again.

How do you adapt a prompt for your trade, your town, and your voice?

Change one layer at a time, and test on a real past job before you trust a new version. Start with the trade layer: swap the bracketed trade, the job types, and the words your customers use, because a plumber's emergency call and a landscaper's spring cleanup should not produce the same first sentence. Next, the local layer: your city, your service area, your hours, and your callback promise, because a prompt that says we reply fast means nothing until the promise has a number on it. Last, the voice layer: read the output out loud, and anywhere it sounds like a brochure instead of your office, replace that line in the prompt itself with an instruction in your own words, such as keep sentences short or never say valued customer.

Keep a before and after example for each prompt you rely on: one real set of notes, and the message you actually sent after editing. When a prompt drifts, and they do drift as models update, that pair shows you in one minute what changed. Save the working versions where your team can reach them, not in one person's chat history. That storage problem is exactly what prompt tools solve, and it is why this directory built the AI Prompt Tools category in the first place.

Here is how the tools in that category fit a contractor office, based on their listing pages in this directory. PromptBase is a marketplace for buying and selling prompts, for an owner who would rather start from a proven prompt than a blank page, with a free section to try the format first. FlowGPT is a community driven platform for sharing and discovering prompts, organized around everyday jobs including Business and Marketing categories, useful when you want to see how other operators phrase the same request. Snack Prompt is a prompt marketplace with a browsable catalog across more than 100 topic categories and a Chrome extension that puts its library inside ChatGPT, which suits an office that lives in the chat window all day. AIPRM is a browser extension layered on ChatGPT with a public prompt library and a private library for your own prompts, organized around jobs such as Marketing, Sales, and Customer Support, which maps closely to the lead, estimate, and review work on this page. PromptHero is a search engine for prompts focused on image and video generation, the stop to make when the job is a project photo treatment or a short video prompt rather than a customer message.

You do not need all of them. Most small offices need one place to save their own proven prompts, and one place to look when a new job appears and no template exists yet. Pick the pair that matches how your office already works, load the eight prompts from this page into it, and edit them there as your wording improves.

What should you check before any AI drafted message goes to a customer?

Check four things, every time, in this order. Facts first: names, dates, times, amounts, and addresses against the source notes, because a fluent message with a wrong arrival window is worse than a plain message with the right one. Promises second: anything the message commits you to, such as a callback time, a start date, or a warranty term, must be a promise the business actually keeps. Price third: any number in the message must trace to your pricing sheet or the invoice itself, never to the model's sense of what a job should cost. Voice last: read it once as the customer, and cut any line you would not say out loud on their doorstep.

That check takes about a minute per message, and it is the real system on this page. The prompts do the drafting. Your office keeps the facts, the promises, and the final say. Businesses that hold that line get the speed the research points to without handing their customer relationships to a model. Businesses that skip the check eventually send one confident, wrong message, and spend a week earning trust back.

Sources named on this page, checked November 13, 2026: RevCore Pro, Speed to Lead for Home Services, summarizing Jobber's 2026 Home Service Trends Report (survey of 1,050 home service business owners, December 2025); 360 Media Consulting, How to Turn More Estimates Into Paying Customers, on qualifying before pricing, clear scope over bare numbers, and scheduled estimate follow up; Contractors Marketing Pros, Contractor Reputation Management, on asking at handover and the final walkthrough, direct review links, and genuine reviews only; and the listing pages for PromptBase, FlowGPT, Snack Prompt, AIPRM, and PromptHero in this directory's AI Prompt Tools category, for what each prompt tool does. Demand evidence for this topic is a search proxy, based on the contractor lead follow up, missed call, estimate, and review request guides ranking for these jobs, not a direct Google Trends read.

Frequently Asked Questions

Can I use these prompts in ChatGPT, Claude, or Gemini without changing them?

Yes. Every prompt on this page is plain text with bracketed fields, so it runs as written in ChatGPT, Claude, Gemini, or a prompt tool extension. Fill in every bracket before you send it, because a leftover bracket is the model telling you a fact is missing. If one model writes longer answers than you want, add a word limit line to the prompt rather than editing every output by hand.

What should I never paste into a prompt?

Never paste anything you would not want stored outside your office: full payment details, access codes you have not been given permission to share, or private customer information beyond what the message needs. Names, job details, and dates are usually enough to draft a message. When in doubt, draft with initials and add the full details yourself in the final message.

How many of these prompts does a small contractor office actually need?

Start with three: the new lead reply, the missed call text back, and the review request. Those three cover the start and the end of most jobs, and they are the messages where speed and consistency pay back fastest. Add the estimate, follow up, reminder, invoice, and update prompts one at a time as each new template proves itself on real jobs.

Will AI drafted messages sound fake to my customers?

They sound fake when the prompt is generic and the output goes out unchecked. They sound like your office when the prompt carries your facts, your times, and your sender's name, and a person reads the draft once before sending. Keep sentences short, name real people, and cut any line you would not say on a doorstep.

Do I need to buy prompts, or are the free ones on this page enough?

The eight prompts on this page cover the core customer messages for most home services businesses, and they are free to copy and adapt. Marketplaces such as PromptBase and libraries such as FlowGPT and Snack Prompt, all listed in this directory's AI Prompt Tools category, earn their place when you need a prompt for a job this page does not cover, or when you want to compare your wording against prompts other operators already use.

How do I keep my prompt library from turning into a mess?

Save prompts in one place your whole office can reach, name each one by the job it does, and keep one before and after example with it: the notes you pasted in and the message you actually sent. Review the library once a season, delete versions nobody uses, and update the callback times, hours, and sender names in the templates before busy season starts.