Be the name the assistant gives at 7am.

Yes, and we confirm it before any budget is committed. OpenAI treats financial services, healthcare and medicine, and legal services as restricted categories approved case by case, and Search Engine Journal has tracked how those approvals work in practice. Home services sit outside that net, so a repair and installation business faces the ordinary review rather than a special one. We still check your specific offers and claims first, and the answer-layer work proceeds either way.

That is exactly the comparison we hold it to. The median LSA cost per lead across 24 home service trades in 2026 is about $40, and garage door leads specifically sit around $15 to $50 because emergency urgency meets moderate ticket sizes. We set the target from your own numbers, not the benchmark, then report cost per booked job side by side across channels so the budget follows the evidence rather than the newest platform.

Both, and we separate them from the first week because they behave differently. Emergency work closes on the call, so speed of answer and a credible response promise decide it. Installation quotes take days and several touches, so the proof on your pages does more of the work. Running them as one campaign hides which is paying, which is the most common problem we inherit. Split apart, most operators discover their real margin sits in one of the two and can fund it properly.

Usually we improve it in parallel rather than waiting, and that is the reassuring part: nothing has to be rebuilt from scratch before you see movement. An assistant needs something specific to quote, so the pages that describe your repairs, brands, coverage and response times get rewritten first while the rest of the site stays exactly where it is. If the site is genuinely holding you back, we say so early and scope that separately instead of quietly charging for it.

Heavily, because reviews and listings are the most quotable evidence about a local business. With 84% of homeowners searching before they choose a contractor and 76% arriving with no company in mind, the reply is often the shortlist. We build a review cadence into the job close so it keeps running without anyone remembering to ask, keep hours and coverage consistent across profiles, and publish real photos of finished work rather than manufacturer stock.

No, it shapes it. Storm months and cold snaps drive spring and cable failures, while installation demand follows the moving and remodelling season, so we pace budget to those curves instead of spreading it evenly and wondering why autumn underperforms. In the quiet stretches we shift weight to the answer layer and to planned installation demand, which has a longer decision window. The monthly report shows the seasonal shape explicitly so you can staff against it.

Alongside, not instead. Search still produces most emergency demand for the operators we work with, and we would not recommend moving that budget. What is changing is the layer above it: a billion weekly ChatGPT users, a fifth showing commercial intent, now across 31 European markets as well as the US. We usually keep your existing programme funded, add a measured position here, and let cost per booked job decide the balance over the next two quarters.

Campaigns produce data in days, but the numbers only mean something after a few weeks of jobs. Movement in how the assistants describe you usually shows within one to three months of the answer layer going live, because these systems re-read sources faster than classic search settles. A steady flow of booked work from this channel is a six to twelve month proposition, and we agree the leading indicators up front so nobody is guessing in month two.

A fixed monthly fee for the work, quoted after the audit rather than off a rate card, plus whatever media budget you approve and pay directly. Most garage door operators start with the audit and the answer layer, which is the part that keeps paying regardless of platform, then add media once the pages can carry it. We agree a target cost per booked job in month one and report against it, so the spend is easy to defend or easy to stop.

It will, and the plan assumes it. The pages, reviews and proof we build are yours and keep working across every assistant and classic search whatever any one platform decides about advertising. The media side stays deliberately flexible: small tests, monthly commitments, nothing that depends on a single placement type surviving. We read the ad policy pages and changelogs weekly and tell you what changed and what it means for your spend.

Yes, and the reporting gets more honest when one team holds it. Our AI search, paid search and analytics specialists work from one plan, so a booked job is attributed once instead of claimed by three suppliers. Hiring us for this channel alone is equally fine — we document the setup and leave every account open in your name.

About an hour a week once it is running, and a little more in the first fortnight while we learn your trade. We need someone who can answer practical questions — which brands you handle, how fast you really get there, what you charge for a call-out — because that detail is what makes the pages quotable. Everything else, including the review cadence and the campaign work, runs on our side.