How a North Texas HVAC Company Went From #6 to #1 in Local Search

A North Texas HVAC company came to us at 60 Google reviews, a 4.4 average, and ranked #6 in the local pack for “HVAC near me” in their service area. Ninety days later they were at 235 reviews, a 4.7 average, and the #1 result on the map. This is the actual four-phase playbook we ran for them — not a hypothetical, the operational sequence from week one through week twelve. Name withheld at the client’s request.

The starting point

The company was technically solid — experienced techs, fast response times, fair pricing. The problem was visibility. In the DFW HVAC market they were competing against three established competitors who sat above them in the map pack, all with review counts between 180 and 400. Even when customers searched specifically in this company’s coverage area, the three competitors were getting the first call.

Phase 1 — setup (week 1)

The first week was about getting the infrastructure in place before sending a single request. We connected their Google Business Profile, pulled their last 12 months of completed-job data from ServiceTitan via API, and tagged the export by service type (install, repair, maintenance) and ticket size. That tagging mattered later when we built the message templates.

Then we wrote the SMS and email templates. HVAC has its own vocabulary, and a generic template (“Thanks for choosing us!”) underperforms a category-specific one (“Thanks for letting us out to look at your AC today”). We wrote three SMS variants and three email variants, each tuned to install vs. repair vs. maintenance, and got owner sign-off on the wording before anything went out.

Phase 2 — new-customer trigger (weeks 2–4)

Starting week two, every completed job dispatched an automated review request. The SMS fired exactly two hours after the technician marked the ticket complete in ServiceTitan, with an email follow-up at day three for non-responders. Both messages routed customers through a thumbs-up / thumbs-down sentiment check first — happy customers went straight to the Google review form, unhappy customers went to a private form that pinged the owner within minutes.

The two-hour delay was deliberate. We tested same-day immediate sends and got lower response rates — customers were often still on the phone with their family talking about the repair, not in the mindset to pick up a review request. Two hours gave them time to settle but kept the experience fresh.

New-customer requests through this phase produced about 35 new reviews in the first three weeks.

Phase 3 — past-customer reactivation (weeks 4–8)

The biggest single source of volume in the program was the 12-month archive. We had roughly 700 past customers with usable contact info, and we dripped requests out at 50 per week over four weeks. Pacing was important — pushing 700 requests in a single day would have produced a one-day review spike that Google’s algorithm is explicitly tuned to flag as suspicious. Fifty per week looked like normal organic growth.

Reactivation messages used a slightly different opening — acknowledging that some time had passed (“You worked with us back in April…”) — which consistently outperforms treating a 6-month-old customer like a same-week one. Phase 3 generated 105 additional reviews on its own.

Phase 4 — reply layer (weeks 8–12)

By week eight the new-review volume was high enough that the owner could no longer keep up with replies manually. We turned on the AI reply layer — every new review got a contextual draft within minutes, our team reviewed and approved each draft, and replies posted within 24 hours of the original review going up. For the one 2-star review that came in during this period (a scheduling miscommunication on an emergency call), the owner handled the reply personally with a direct phone number, called the customer, and the customer voluntarily updated the review to 4 stars two weeks later.

Three operational details worth calling out

Most of the playbook above is recognizable as standard review-program structure. Three specific details mattered more than they look on paper.

The two-hour delay on the SMS. Not 30 minutes, not the next morning. We tested both. Two hours after technician sign-off was the sweet spot — settled enough to be reflective, fresh enough to be specific.

Service-type-specific email variants. A new AC install ($8,000 job) and a $189 capacitor replacement are not the same emotional event for the customer. The email wording for the install was warmer and more reflective (“how was the install experience overall…”), the maintenance email was shorter and more transactional. Same sentiment check, different on-ramp.

The 2-star recovery. A negative review is going to land eventually. What mattered was that the sentiment check on the front end had already filtered out the avoidable negative reviews, so when a real one came through, the owner had the bandwidth to handle it personally and well. The customer-driven update from 2 stars to 4 stars produced a profile that looked better than if the review had never happened.

What this took from the owner

Roughly 90 minutes total during onboarding — sign-off on the message templates, the ServiceTitan API token handoff, and a kickoff call to walk through the four phases. After that, the owner’s involvement was reading the weekly summary and handling the one negative review that came in. The system did the rest.