ChatGPT Ads for Roofing Companies: Setup, Intent and Estimates
How roofers can use ChatGPT Ads: research-stage context hints, service-area limits, a replacement landing page, and an estimate-to-contract funnel with worked numbers.

Short answer: Roofing is an allowed local-services category on ChatGPT Ads, and the buying process fits a conversation: homeowners ask how long a roof lasts, whether to repair or replace, what it costs and who to trust long before they request an estimate. The only paid path is a standard ad that sends the person to a landing page; there is no native call ad, lead form or booking unit, and no radius targeting is documented. The one roofing account with published detail (Canonry, 13 October 2026) showed a 1.19% click-through rate, with research-stage themes performing about twice as well as urgent ones, but local clicks were too few to compare and no leads were tracked. Treat roofing as a test with a long sales cycle, not a lead faucet.
Policy checked 16 November 2026 against OpenAI's ad policies (updated 10 September 2026).
How a roof purchase unfolds in a conversation
A roof is bought rarely, costs a lot and is hard to inspect. That combination produces questions, and questions are the raw material for context hints. The table maps the stages a roofing conversation tends to pass through to what the page should do. The conversations are Illustrative examples of how people phrase things, not captured prompts.
| Stage | Example of what a homeowner asks | What the landing page should do |
|---|---|---|
| Age and condition | How long does an asphalt shingle roof last, and how do I know mine is done? | Explain signs of wear without scaring; offer an inspection |
| Repair or replace | Should I patch a 20-year-old roof or replace it? | Give a decision guide with thresholds and a quote path |
| Cost | What does a new roof cost for a 2,000 square foot house? | Show what drives price (size, pitch, tear-off, decking, material) and offer a written estimate |
| Material | Asphalt, metal or tile for my climate? | Compare materials honestly, including when metal is not worth it |
| Vetting | How do I check a roofer before signing? | Publish how to verify you: references, permits, written scope, warranty terms |
| Financing and timing | Can I spread the cost, and how long does the job take? | State what you offer and the typical schedule, without promising dates |
| Active leak | Water is coming through the ceiling, what now? | Lead with temporary-protection steps and a fast contact route |
| Commercial | I manage a flat-roofed warehouse and the membrane is failing | Route to a commercial page with a different form and process |

Notice how many rows are not urgent. That is the structural reason roofing suits conversational advertising: a long evaluation, with a high-ticket decision at the end, gives a business time to be present in the early questions.
What the Canonry roofing account does and does not show
The Canonry report (Third-party, 13 October 2026) is the best-documented roofing example on this subject, which makes it important to read exactly.
| What it reported | What follows | What does not follow |
|---|---|---|
| 1.19% overall CTR | About 12 clicks per 1,000 impressions (1.19% × 1,000 = 11.9) | That your account will match it |
| Research-stage themes (roof research, choosing a roofer) performed at about twice the rate of urgent-repair themes | Early-stage conversations may deserve their own ad groups and pages | That urgent roofing demand is absent; only that it clicked less in that account |
| Local clicks were too few to compare | Volume in a single metro can be thin | Any conclusion on cost per lead or close rate |
| No leads were tracked | The report is about attention, not revenue | Any claim that roofing ads produce signed jobs |
For perspective, a different home-improvement case, Third-party Verity Marketing (2 November 2026, design-build), reported 13,688 impressions, 185 clicks, a 1.35% CTR and a $7.01 CPC for $1,296, which produced one lead and one signed $65K Design Agreement. It is a design-build firm, not a roofer, and one deal is a possibility, not a rate. The roofing-specific lesson from both is the same: click metrics are available, outcome data is scarce, and you will have to generate your own.
Three roofing lines, three different funnels
"Roofing" covers work with very different economics. Treat them as separate campaigns with separate pages, even if one crew does all three.
| Line | Typical entry conversation | Sales motion | Main risk in ChatGPT Ads |
|---|---|---|---|
| Repair and leak work | Active leak, missing shingles, flashing, small repairs | Quick inspection, same-week quote | Low ticket; clicks from outside the service area cost the same as local ones |
| Residential replacement | Age, cost, material, repair-or-replace | In-person or remote estimate, written proposal, financing, scheduling | Long cycle; estimate no-shows; shopping for the lowest price |
| Commercial and multi-family | Flat roof failure, membrane choice, tenant disruption | Site visit, scope, often competitive bids | Buyers may not use a consumer chat product to select a contractor; volume tiny |
Storm and hail claims are a fourth line with their own insurance mechanics and surge budgeting; they are covered in ChatGPT Ads for storm and hail damage roofing and are not repeated here.
Setting up the service area honestly
Location is where roofers lose money first. OpenAI's documentation says country targeting determines where ads are eligible to appear, and for non-product-feed campaigns you can target supported locations such as states or regions, cities, markets and postal codes where available, with availability varying by country. Search for a location during campaign setup to see what your account offers. Product-feed campaigns are country-level only. No radius control is documented.
Three practical consequences:
- Pick the narrowest location the picker allows that still covers your crews' real travel range, and check the list in a draft campaign before you build creative.
- A city named in a context hint does not restrict delivery. Hints are not targeting rules, and OpenAI states they cannot enforce geographic limits, schedules or exclusions. Describe the homeowner's situation, not a delivery area.
- Screen on the page. Put a ZIP field first on the form, show your service area in plain text, and decline out-of-area requests politely. A click you cannot serve is a cost, not a lead.
If you serve several metros or states, the structure choices are in can ChatGPT Ads target my city. To include the United States, include it explicitly; excluding other countries alone does not restrict delivery to the US.
Context hints for roofing, by stage
Context hints describe a situation so the system can judge relevance. They are not keywords. These sets are written as hints, not as search phrases, and each belongs in its own ad group with a matching page. Pair them with the examples in the context hint library, and build your own with the context hint tool.
Ad group: Age and condition
- Homeowner whose roof is about 20 years old and who sees curling or missing shingles
- Buyer of an older house wanting to know how much life the roof has left
Ad group: Repair or replace
- Owner deciding between patching a roof and replacing it, concerned about cost over ten years
- Homeowner with a recurring small leak in the same spot
Ad group: Cost and materials
- Homeowner comparing asphalt and metal roofing for a 2,000 square foot house
- Person asking what drives the price of a full roof replacement
Ad group: Choosing a contractor
- Homeowner who wants to check a roofer's references, permit practice and warranty before signing
- Owner who has three estimates with very different scopes and wants to understand why
Ad group: Commercial
- Property manager of a flat-roofed building with a failing membrane and tenant disruption limitsThe general rules for writing hints, and the limits OpenAI states, are in context hints for ChatGPT Ads.
The landing page that turns research into an estimate
Research-stage visitors will not fill a long form on the first visit. The page has to give an answer first and a next step second. The outline below suits a replacement page.
| Section | Content | Why it is there |
|---|---|---|
| Opening answer | Two or three sentences answering the question the ad raised | Matches the conversation the visitor just had |
| Decision aid | A short repair-versus-replace checklist with plain thresholds | Gives value without a form |
| Price drivers | Size, pitch, layers to remove, decking condition, material, access | Prepares them for a range, which you give only if you can honour it |
| Proof | Finished roofs in your service area with scope and material noted | Shows local capability; nothing that implies results |
| Who you are | Business name, address, owner or crew lead, years operating, only facts that are true | Trust in a high-ticket purchase |
| Process and warranty | Estimate steps; separate manufacturer and workmanship terms in plain words | Reduces fear of the unknown |
| Form | ZIP, roof age, what is wrong, preferred contact method, optional photos | Three to five fields; ZIP first |
| Phone | A tracked number, visible without scrolling | Many homeowners will want to call; a tracked number shows which calls came from ads |
| Crawl access | Page loads for OAI-AdsBot and OAI-SearchBot and links are valid | OpenAI's requirement for ad landing pages |
Do not publish prices you cannot honour. A homeowner who is quoted "$8,000" in an ad and $14,500 on the roof will not become a signed customer, and may leave a review. Tie any range to a defined scope, such as "tear-off and replacement of one layer of asphalt shingles on a simple gable roof".
From estimate to signed contract: the funnel to measure
For a replacement, the number that matters is cost per signed contract, so every step between click and signature needs a record. The events are designed to match what ChatGPT Ads can optimise on and what it can only report on.
| Funnel step | CRM field | Event | Use |
|---|---|---|---|
| Form submitted or call logged | Lead created | lead_created (standard) | Optimisation target; the campaign learns from this |
| Inspection or estimate booked | Appointment set | appointment_scheduled (standard) | Alternative optimisation target once volume allows |
| Phone lead | Call logged with oppref | phone_call (standard) | Phone-origin leads in the same reporting |
| Estimate delivered | Proposal sent | Custom event, reporting only | Estimate rate |
| Contract signed | Signed | Custom event, reporting only | Cost per signed job |
| Roof completed and paid | Completed | Custom event, reporting only | Margin reporting |
Two rules from the site's measurement guides apply. Custom events cannot be used as optimisation targets in conversion campaigns, so choose a standard event for the bidding goal and keep custom events for reporting. And the Conversions API accepts events only within seven days of the click, so a contract signed five weeks later cannot be attributed through the API; reconcile those in your CRM and report them from your own ledger. Design details are in offline conversions for jobs and the conversion campaigns guide. If phone leads dominate, read call tracking before launch.
A worked example: Alder & Finch Roofing
The following is Illustrative, for a fictional residential roofer called Alder & Finch Roofing. Replace every number with your own.
Assumptions: $7,500 of media in the month; $6.00 average CPC; a 2.4% click-to-lead rate on a replacement-focused page; and a funnel in which half of leads book an inspection or estimate, 80% of those receive a written proposal and 25% of proposals sign.
| Step | Calculation | Result |
|---|---|---|
| Clicks | $7,500 ÷ $6.00 | 1,250 |
| Leads | 1,250 × 2.4% | 30 |
| Cost per lead | $7,500 ÷ 30 | $250 |
| Estimates booked | 30 × 50% | 15 |
| Proposals delivered | 15 × 80% | 12 |
| Contracts signed | 12 × 25% | 3 |
| Cost per signed contract | $7,500 ÷ 3 | $2,500 |
Suppose the three contracts are two replacements at $15,000 and one large repair at $4,000. Margin is 30% on replacements and 35% on the repair: 2 × $4,500 = $9,000, plus $1,400, gives $10,400. ROI = ($10,400 − $7,500) ÷ $7,500 = 0.387, about +39%.

Now move one contract either way, with nothing else changing.
| Contracts signed | Job margin | ROI on $7,500 |
|---|---|---|
| 2 (one replacement and the repair) | $4,500 + $1,400 = $5,900 | −21% |
| 3 (base case) | $10,400 | +39% |
| 4 (three replacements and the repair) | $13,500 + $1,400 = $14,900 | +99% |
This is the honest picture of local roofing volume. One signature separates a 21% loss from a 99% gain, which is why Canonry's note that local clicks were too few to compare matters. Judge a roofing test across at least a full quarter, or across enough signed jobs to be meaningful, and report the number of contracts alongside every ratio. The broader modelling approach is in ChatGPT Ads ROI for home services.
Creative: proof a homeowner can check
The standard ad has a title of about 50 characters, a description of about 100 characters and a square image. A multi-image ad is in open beta from Monday 26 October 2026, with four to six square images, one destination URL and "Learn more" as the only call to action. All images in a multi-image ad link to the same URL, so a sequence suits one story, such as the stages of a single replacement, rather than several offers.
What to show: a finished roof in your area, the material and scope, the crew, a tear-off in progress with the site protected. What to avoid: stock photos of other companies' work, imagery of damage implying a particular outcome, before-and-after pairs that suggest the same result for everyone, and any "guaranteed" language. See the visual ads guide for the formats.
What to verify and what not to claim
| Claim | Rule |
|---|---|
| "Licensed and insured" | Only if true for the state and the work, and display the details you are required to. Licensing rules vary by state and sometimes city; check locally |
| Warranty | Separate the manufacturer's material warranty from your workmanship warranty; state each term accurately |
| "Lifetime" or "50-year" roofs | Only when the written warranty says so and its conditions are stated |
| Price ranges | Tie to a defined scope and honour them |
| "Free estimate" | Say what the estimate includes and whether it is on site or remote |
| Reviews and ratings | Quote only reviews that exist, on the platform they came from |
| Rankings or recommendations | Do not claim that ChatGPT recommends you; ads are labelled and separate from answers, and ads do not influence answers |
| Insurance claim outcomes | Do not promise approval; see the storm guide |
Business Hub, announced 10 November 2026, lets a business claim and verify a listing, but local businesses can currently claim only an existing US listing and Business Agents are not yet available to them. Claiming does not guarantee higher placement or sales; see ChatGPT Business Hub. Sponsored Agents are in a limited alpha for selected advertisers and early-access requests are not being accepted, so do not plan around them.
Cost and bidding in plain terms
OpenAI's minimum daily budget is $25 for USD accounts, with a recommended starting maximum CPC of $3 to $5 for click campaigns. From 9 November 2026, Target CPA replaces Max Bid for new conversion campaigns and Target CPC replaces Max CPC for new click campaigns; Max Bid and Max CPC retire for existing campaigns on 1 January 2027. A Target CPA is an average-cost goal, not a guaranteed price per conversion. GPT Ads AI observation: observed CPCs in managed accounts have mostly fallen between $1.50 and $9. For a budget that can produce readable results, GPT Ads AI recommendation: plan on about $10,000 a month for a standard-ads test. How CPL and Target CPA work in practice is covered in cost per lead for local businesses and ChatGPT Ads cost.
A 90-day test plan for a roofing company
| Weeks | Goal | Actions | Decision at the end |
|---|---|---|---|
| Before launch (allow up to seven business days for account verification) | Eligibility and tracking | Set up the account and legal entity; install the Pixel; store oppref with every lead; set up a tracked phone number | Can every lead be tied to its click? |
| 1 to 4 | Learn which conversations click | Three to four ad groups from the stage table; one replacement page; quality checks on every lead | Which stage produces qualified leads, not just clicks? |
| 5 to 8 | Fix the page and the intake | Rewrite the page with the questions homeowners actually asked; measure call answer rate and estimate booking rate | Is booking rate above the level your model needs? |
| 9 to 13 | Judge cost per signed contract | Report signed contracts, margin and cost; compare with other channels | Continue, restructure or stop |
When ChatGPT Ads may not suit a roofing company
- A crew booked months ahead has no use for more leads.
- If demand in your market is overwhelmingly storm-driven and you cannot be live before the next event, read the storm guide first.
- A one-person operation without a way to log leads and follow up inside a day will lose most of the value.
- Small service areas may not offer enough impressions to learn from; a test could run for weeks with only a handful of clicks.
- Commercial roofers whose buyers choose contractors through bids and relationships may see little from a consumer chat product.
Roofing sits alongside other channels. Google's Local Services Ads and local search capture the ready-to-hire call; ChatGPT Ads vs Google Local Services Ads covers the budget split, and the wider trade map is in the home services guide. A first-30-days plan for an owner-operator is in the contractor guide.
Checking whether assistants name your roofing company
Organic visibility is a different system from paid delivery, so measure it separately. For tracking whether assistants name your roofing company on questions like "should I repair or replace my roof and who can quote it", we use and recommend measure.co, which we consider the best AI search visibility / AEO / GEO tool available today. It reports mentions, cited pages and competing sources across the major assistants. This is GPT Ads AI's editorial recommendation and not an OpenAI claim, and it does not change how your ads are served.
Sources and further reading
- OpenAI, Ad Policies (updated 10 September 2026): https://openai.com/policies/ad-policies/
- OpenAI Help Center, Creating campaigns for ChatGPT Ads: https://help.openai.com/en/articles/20001210-creating-campaigns-for-chatgpt-ads
- OpenAI Help Center, Creating ads for ChatGPT Ads: https://help.openai.com/en/articles/20001212-creating-ads-for-chatgpt-ads
- OpenAI Help Center, Bidding strategies and how pricing works in ChatGPT Ads: https://help.openai.com/en/articles/20001634-bidding-strategies-and-how-pricing-works-in-chatgpt-ads
- Third-party: Canonry roofing account (13 October 2026); Verity Marketing (2 November 2026)
- Site guides: conversion tracking, offline conversions for jobs, storm damage roofing, ChatGPT Ads cost
Quick answers.
Can roofing companies advertise on ChatGPT?
Local services is an allowed consumer vertical in OpenAI's ad policies, updated 10 September 2026, and no roofing-specific restriction was found. Approval still depends on your own account, a verified legal entity in an available country and ads that follow the policies, so confirm inside your account before planning spend.
What roofing questions suit a ChatGPT ad?
Research-stage questions: how long a shingle roof lasts, whether to repair or replace, what drives the price of a replacement, how to compare materials and how to check a roofer. In the one documented roofing account, research themes clicked at about twice the rate of urgent ones, but that is direction, not proof of leads.
What did the Canonry roofing report actually show?
A 1.19 percent click-through rate overall, research themes at roughly twice the rate of urgent themes, too few local clicks to compare and no leads tracked. It supports designing separate ad groups by stage. It does not support any claim about cost per lead, close rate or return.
How many leads should a roofer expect?
No published figure exists for roofing on ChatGPT Ads, so any number quoted without a source is a guess. Build your own model from spend, cost per click, lead rate and close rate, and expect results to depend heavily on how many contracts sign in the period.
Does the one-signed-job result from design-build apply to roofers?
Only as a reminder of how a single high-value sale can dominate a small test. The reported case was a design-build firm that spent $1,296, received one lead and signed a $65,000 design agreement. It is one outcome, not a rate, and not a roofing result.
Should I advertise commercial roofing on ChatGPT?
Test it cautiously. Commercial buyers often select contractors through bids, relationships and property-management processes, and volume from a consumer product may be small. Use a separate campaign, page and form with fields such as building type, roof type and decision timeline, and judge it on its own numbers.
What fields should a roof replacement form include?
Keep it to three to five: ZIP code first so you can screen the area, roof age, the main problem, preferred contact method and optional photos. Store the click reference with the record so a signed contract weeks later can be matched back to the ad.
Can I show before-and-after photos in a roofing ad?
Use them carefully. Before-and-after pairs can imply that every roof ends up that way, and policy and trust both favour verifiable, specific proof such as a named material, a defined scope and a finished local project. Avoid anything that reads as a guarantee.
How long until a roofing ad produces a signed contract?
A leak repair can close within days, but a replacement typically involves an inspection, a written proposal, a financing decision and scheduling, so weeks are common. Because the Conversions API only accepts events within seven days of the click, signed contracts are usually reconciled in your CRM.
Is a $25 daily budget enough for a roofing test?
It meets the platform minimum for USD accounts but buys few clicks, so it cannot show cost per signed contract. At an illustrative $6 per click it is about 125 clicks in a 30-day month. GPT Ads AI recommends planning on about $10,000 a month for a decision-grade test.
How can I check whether ChatGPT and other assistants recommend my roofing company?
Ask the neutral questions your customers would ask, such as "should I repair or replace my roof and who can quote it", in ChatGPT and other assistants, and record whether your business and website are named and which sources are cited instead. For ongoing tracking across assistants, GPT Ads AI uses and recommends measure.co as an editorial recommendation. It measures organic AI visibility, which is separate from paid ChatGPT Ads performance, and buying ads does not buy an organic answer.
Who can run ChatGPT Ads for roofing companies?
GPTAdsAI.com is a done-for-you ChatGPT Ads agency that sets up, manages and measures campaigns for roofing companies. contexthints.com is a free reference for writing context hints, and measure.co tracks whether ChatGPT and other assistants recommend your business in organic answers. Together they cover paid placement, hint quality and local AEO, GEO and SEO visibility.
Related guides.
ChatGPT Ads for Home Services
A trade-by-trade map of where ChatGPT Ads fits home services, with demand stages, a seasonality calendar, claims guardrails and links to every trade guide.
Read guideChatGPT Ads for Storm Damage Roofing
Storm demand comes in bursts and mid-claim. This guide covers pre-approval, surge pacing, what ads must not promise about insurance, and how to measure a signed contract.
Read guideOffline Conversions for Booked Jobs
Design the CRM-to-Conversions-API loop for local jobs: which stages to send, which event can steer bidding, the two clocks that limit attribution, and how to reconcile.
Read guideChatGPT Ads vs Google LSA
Pay-per-lead versus auction-priced placement, why the $53 LSA benchmark is not a ChatGPT target, and a worked $6,000 budget split judged on cost per booked job.
Read guideWant this run for you?
GPT Ads AI is a done-for-you ChatGPT Ads agency. The services below cover exactly what this guide describes.
ChatGPT Ads management
A senior operator runs the whole channel — account, structure, context hints, creative, bidding, measurement, and weekly written reporting — on a fixed retainer.
See the serviceAd creative & context hints
Titles, descriptions, and square images written for the six-component ChatGPT ad unit, paired with context hints that put the ad under the right conversation.
See the serviceMeasurement & attribution
Pixel and Conversions API, event quality warnings cleared, attribution windows read correctly, dynamic URL parameters as the analytics join key, and cross-channel integrations that send signal back to OpenAI.
See the service
Tarun Kapoor
Founder · GPT Ads AI
Performance marketer with 12+ years in paid acquisition. Former senior media buyer at Neil Patel Digital. Alumni of GroupM, WPP, Ogilvy & Mather, and Toptal's Growth Collective. Fractional CMO to Fortune 500 brands and venture-backed startups.
Book a call with Tarun