ChatGPT ads for contractors.

Contractors buy estimates, not clicks

A contractor campaign should be designed around qualified estimates and booked jobs, not cheap clicks. The value of an estimate depends on the trade, job type, service area, timing, gross contribution, close rate, and crew capacity, so one campaign structure cannot serve every contractor equally well. A restoration dispatch, an electrical diagnostic, and a remodeling consultation represent different commitments for both the homeowner and the business.

Start by defining what the business is prepared to accept. For each service line, document the problem served, minimum job fit, geography, property types, hours, required licenses, estimator or technician owner, and the event that counts as a real opportunity. This definition should control the ad promise, landing page, intake questions, routing, follow-up, and CRM stage. If those pieces describe different offers, campaign data cannot identify what is working.

Set bidding posture from estimate economics

A higher-value job can tolerate a higher cost per estimate when its expected gross contribution and close probability support the spend. That is not permission to bid aggressively based on revenue alone. Materials, subcontractors, travel, sales time, warranty exposure, cancellations, financing friction, and production capacity all affect what an acquired job is worth. Use the company’s own completed-job records rather than a generic trade benchmark.

Work backward without pretending the inputs are certain. Estimate the gross contribution of the job type, multiply it by a defensible close probability for genuinely qualified estimates, then reserve room for overhead, operational risk, and profit. The remainder provides a ceiling for acquiring an estimate. Connect that ceiling to observed rates from valid clicks to qualified estimates, and reduce exposure when the sample is immature or sales dispositions are incomplete.

ChatGPT campaign objectives are currently CPM, CPC, and oCPC, which is conversion-optimized cost per click. For CPC campaigns, OpenAI recommends starting with a maximum bid of $3 to $5 USD per click, set at the ad-group level. An oCPC ad group instead uses a Bid Cap representing the most you will bid for a conversion, and OpenAI publishes no recommended amount for it. oCPC steers delivery toward a tracked conversion event, but billing stays per valid click rather than per completed conversion, so treat published guidance as an auction input rather than proof of affordability. Selection uses a relevance-weighted, second-price auction, while contractor economics determine the acceptable limit.

The current minimum daily budget is $25 for accounts billed in USD, and other billing currencies have their own documented minimums. That figure is the platform floor required for a campaign to run, not a recommended contractor test budget. A trade with a high average job value and a long decision cycle usually needs considerably more than the floor before enough qualified estimates exist to judge anything, so size the test from job economics and confirm the current requirement inside your own Ads Manager account.

Give every trade a distinct campaign treatment

Trade-specific architecture starts with the customer decision, not a broad contractor label. Emergency work needs immediate routing and clear availability. Planned replacement needs education, scheduling, and a consultation or inspection offer. Diagnostic work must explain that a technician identifies the cause before a reliable quote is possible. Longer projects require patient follow-up and stronger screening before an estimator invests time.

Split these motions even when one company performs all of them. Their urgency, qualification criteria, landing content, staffing, and booked-job definitions differ. Separate ad groups also let the maximum bid and context hints reflect the real service. Combining them can send an emergency request into a consultation queue or make a major-project team chase minor repair inquiries that the campaign was never designed to serve.

Separate ad groups by job type

Context hints are configured at the ad-group level. They describe relevant conversations, topics, or keywords, but they are not exact-match keywords and do not guarantee delivery in any specific conversation. A coherent ad group gives the system aligned evidence: one job type, a matching title and description, an appropriate landing page, and context hints that describe the same customer need.

Do not place drain clearing, repiping, water-heater replacement, and bathroom remodeling under one plumbing hint set merely because one license covers them. Their urgency and customer decisions differ. The same rule applies to storm roofing and planned replacement, or an HVAC breakdown and a system consultation. A clean split makes lead review interpretable because the team knows which promise and intake path produced the opportunity.

• One customer problem

  Name the symptom or planned project clearly enough that intake can recognize the intended job.

• One offer and destination

  Send the ad to the inspection, diagnostic, emergency call, or consultation path promised in the creative.

• One operating owner

  Assign the queue to the team that can answer, qualify, schedule, and disposition that job type.

• One economic rule

  Set bidding and pacing from the job type’s estimate value rather than a blended company average.

Qualify before consuming estimator time

A free quote can attract people who want remote pricing without enough information, are outside the service area, need unsupported work, do not control the property, or are collecting speculative numbers for an undefined project. Qualification should not create a hostile form. It should ask the few questions needed to route the right next step and explain why an inspection, diagnostic, or consultation is necessary.

Match questions to the trade. Restoration needs incident timing, loss type, location, and immediate safety routing. Roofing needs property type, issue or project goal, storm context, and inspection availability. Remodeling needs ownership, project category, readiness, location, intended timing, and decision-maker participation. Electrical and mechanical trades need symptoms and equipment context, while avoiding a promise that the cause can be diagnosed from a short form.

• Fit

  What service is needed, and is it a job the company currently performs?

• Place

  Where is the property, and can the assigned crew serve it profitably and lawfully?

• Authority

  Does the requester own, manage, or have authority to approve work at the property?

• Timing

  Is this an emergency, a planned decision, or an early research request that needs a different follow-up path?

Protect estimator and crew capacity

Advertising can create an operational problem when estimates arrive faster than the business can inspect, quote, schedule, or fulfill them. Before launch, map capacity by trade, job type, geography, day, and owner. Include emergency reserve, existing backlog, drive time, estimator preparation, proposal work, and the crew’s production calendar. A calendar opening is not useful if the required technician, equipment, or license is unavailable.

Create pacing rules tied to real operations. Route urgent jobs only during genuinely monitored periods, cap consultation slots, suppress services with excessive backlog, and pause areas that create unproductive travel. Review missed calls, uncontacted forms, appointment delay, estimate aging, and production availability alongside campaign delivery. A campaign should not keep accelerating because click metrics look healthy while homeowners wait without a credible next step.

Make booked jobs the decision metric

Ads Manager Beta can report impressions, clicks, spend, click-through rate, average CPC, average CPM, and conversions. Those measures describe delivery and tracked actions, but a contractor still needs CRM outcomes to understand commercial value. Configure available conversion measurement and retain static UTM parameters so campaign visits can be read in existing analytics, then connect each call or form to a consistent service, qualification, estimate, and sales record.

Compare each trade on cost per qualified estimate and cost per booked job, with enough time for its normal decision cycle. Keep emergency repairs separate from long-consideration projects so quick outcomes do not disguise an immature remodeling cohort. Revenue can add context when records are complete, but booked jobs remain the clean operating checkpoint: the customer committed, the business accepted the work, and capacity was assigned.