A med spa we work with was quietly bleeding time on something nobody talks about. Their front desk manager was spending roughly eight hours every week responding to Google reviews. Not writing them, just responding. Reading each one, drafting a thoughtful reply, matching the brand voice, escalating the angry ones to the owner, and posting the response. Eight hours. Every week. For a task that, in theory, takes thirty seconds per review.
We built them a review response system that drafts replies on brand, queues them for one person to approve, and posts on a click. The manager now spends about five minutes a week on the entire workflow. Here is exactly what we did, what we tuned, and what changed.
Before: the hidden cost of being responsive
The owner of this med spa, call her J, takes reviews seriously. Her front desk reads every Google review out loud at the Monday team meeting. She believes (correctly) that responding publicly to every review, positive or negative, builds trust with prospective clients who are reading them before booking.
The problem was operational. Her front desk manager was the only person on staff with the writing chops to respond well, so the task got dumped on her. She would batch the work on Friday afternoons, sit there for two to three hours, and grind through forty to sixty reviews per week. By the end of the month she was burned out on a task that, on paper, had nothing to do with her job description.
The math was uglier than the time spent. That manager makes real money. Her hourly value as a scheduling and intake lead is meaningfully higher than what a writer would charge for the same drafts. The med spa was spending the equivalent of a part time content role on a task that gave them zero new revenue. It just preserved their reputation.
After: five minutes a week, same quality, more reviews handled
We built a system that pulls in every new Google review as it lands, drafts a response in the spa's brand voice, and queues it for one click approval. The manager opens the queue once a week, scans the drafts, edits maybe one or two, approves the rest, and that is the end of her review response job. From three hours of grinding to five minutes of skimming.
The drafts are not generic. They name the treatment when the reviewer mentioned one. They reference the specific provider by first name when the reviewer thanked someone by name. They thank repeat clients by acknowledging it is not their first visit. They stay on the right side of HIPAA by never confirming or implying anyone was actually a patient. None of that is hard once you set the rules clearly, but no off the shelf review tool we tested got it right out of the box.
The rules we tuned
The first week we ran the system in shadow mode. It drafted responses, the manager rewrote half of them, and we logged every edit. After about ten days of edits, we had a clear picture of where the drafts were missing the brand voice. Then we encoded those preferences as rules.
A few of the rules that mattered most:
Never apologize for things that are not the spa's fault. A common pattern in negative reviews is people complaining about parking, weather, or another tenant in the building. Generic AI responses tend to apologize for everything to be safe. The owner hated this. The rule now is to acknowledge the frustration without taking blame that does not belong to the business.
Escalate any review under four stars to the owner. Critical reviews never get auto drafted past a certain confidence threshold. They route to the owner with a suggested response and a one paragraph summary of what the reviewer is upset about. She decides whether to reply, call the client, or both.
Mirror the reviewer's energy, not the reviewer's length. Long enthusiastic reviews do not need a long response. Short reviews do not need a robotic two sentence template. The system writes in the rough register of the reviewer, which sounds more human.
No promotional language in responses. Early drafts had a habit of trying to upsell, "we hope you'll come back and try our hydrafacial next time." The owner banned it. Responses thank, acknowledge, and close. They do not pitch.
What the spa got back
A few weeks in, the manager pulled an extra eight hours of capacity back into actual revenue work: confirming appointments, recovering no shows, calling membership prospects. The owner stopped getting Sunday night texts that started with "I didn't get to the reviews this week." The response rate on Google went from about 60% (her honest pre system number) to 100%, because the cost of replying dropped to near zero.
The less obvious win was consistency. When a human is rushing through forty drafts on a Friday, the last ten get sloppier than the first ten. The system does not get tired. The fortieth response is as on brand as the first.
What we did not change
We did not automate the part that matters most: the human reading the drafts. The manager still approves every single response before it posts. The owner still reads every negative review personally. The system did not replace human judgment. It replaced human typing. That distinction matters, and it is the same one we draw for every client we build review or response automation for.
If you are running a service business and your team is spending real hours on something that is mostly pattern matching, that is usually a sign there is a system to be built. We offer a free discovery call where we walk through where your team is spending time, what is repeatable, and what is worth building. If you want to see what your version of this looks like, that is the place to start.