Your Follow-Up Automation Is On. It's Also Not Working.

Six months. Volume growing, trucks staying busy, booking rate solid. And my Google review count sat between four and six new reviews per month like it was bolted to the floor. Same number in March as September. NiceJob sending the requests. Housecall Pro firing the completion texts. Everything configured.

Nothing working.

That was 2023. I ran an audit. What I found is the whole reason this article exists.

---

What "Automated Follow-Up" Actually Means at a 5-Truck Shop

Here's what I had running: Housecall Pro handling the completion message and invoice, NiceJob picking up the review request the next day. Built it in a Saturday afternoon in early 2022 and told myself the problem was solved.

What I didn't do was go back and check whether it was working. "Configured" and "working" are not the same thing. I treated them like they were for almost a year.

The 2023 audit broke that assumption. I pulled review velocity alongside job volume for the previous six months. The mismatch was obvious. Volume was up. Reviews per month hadn't moved. The automation was running on schedule. Nobody was responding to it, or nobody was reading it, and I had no idea which because I hadn't looked once.

---

The Real Problem: Your Triggers Are Dumb and Your Timing Is Wrong

The architecture most shops are running is time-based. The clock hits 24 hours after job close, the message goes out. That's the entire logic.

It doesn't know what happened on the job.

One of my techs red-tagged a Federal Pacific panel on an EV charger pre-install. Customer had called us to run a 50-amp circuit for a Tesla. My tech got in the panel, saw the FPE breakers, and had the conversation nobody wants to have — your panel is a fire hazard, we can't proceed, here's what a replacement looks like. That's exactly what I want my tech doing.

Twenty-four hours later, NiceJob sent that customer a review request.

One star. Not because my tech did anything wrong. Because they were still sitting with news they hadn't budgeted for — a $4,200 panel replacement — and our cheerful automated message hit them at exactly the wrong moment. The system didn't know any of that. It just knew the clock had expired.

I talk constantly about call recording because the discipline of listening back to what actually happened is the same discipline you need for follow-up. You set the automation up, you assume it's running, you never look at what it's producing. That's not a software problem. It's a looking problem.

---

More Automation Is Not the Answer

The standard contractor marketing advice is to build a longer sequence. Seven touchpoints. A re-engagement campaign at 30 days. A seasonal check-in at 90.

I'm going the other direction.

A broken follow-up sequence doesn't get better by adding steps. It gets better with fewer automated messages and one human phone call placed in the right spot — 48 hours after job close, on any job that didn't close clean. A dispatcher picking up the phone and calling the customer to ask how things are going.

The objection I hear: it takes four minutes of dispatcher time per call. At volume, that adds up. The 7-step sequence is cheaper.

Here's what I know about phone calls at the right moment: my NPS at Reeves Electric was a 4 in month two. By month nine it was an 81. I didn't change my pricing. I didn't launch a new service line. I answered every call myself for 90 days and personally followed up on every job I had questions about. A real voice closed loops that texts never touched.

AI is going to eat this workflow. I think we have 36 months before it's genuinely good enough to hand off, and most shops will be late to that transition. But right now, the human moment that matters most is not where most shops are putting it. They're automating the 48-hour call to save four minutes.

That's the wrong trade.

---

The Audit I Actually Did

In Q1 2024, I sat down with my dispatcher on a Thursday morning and pulled the automated messages that had fired the previous quarter. We cross-referenced two things: whether the job produced a review or a callback complaint, and whether the customer booked again within 90 days.

One cut was hard to ignore. We'd flagged 23 jobs that quarter as "needs follow-up" — jobs where my tech had written a note, or my dispatcher had made a mental note, or someone flagged a complexity at close. Every one of those 23 jobs still received the standard completion text and the 24-hour review request. Same template as a clean panel swap.

Most of them produced nothing. The ones that produced something were jobs where my tech had happened to call the customer personally before leaving the driveway — not because of any system, just because he felt like something was unresolved.

After that audit, my dispatcher and I started a weekly follow-up review. Same format as the call recording sessions I already run. We pull a handful of sequences, tag what the trigger logic should have done differently, and close the tab. It takes 25 minutes. It's now a standing Thursday item.

The tool runs whether you check it or not. The cadence forces you to look.

---

What the Smarter Architecture Actually Looks Like

The core change: my techs now update a status field when they close a job in Housecall Pro. The options are "clean," "callback likely," "red tag issued," and "customer had questions." That field runs through a Zapier logic branch. Clean close goes to one sequence. Anything else routes to a separate track that suppresses the automated review request and drops a task in my dispatcher's queue for a 48-hour phone call.

NiceJob only fires on clean closes. And I moved the timing from 24 hours to 72.

That timing change alone made a difference I didn't expect. Customers had more time to actually experience the work before we asked them to comment on it. The quality of reviews went up. The volume went up with it.

Before the rebuild: four to six new Google reviews per month. After two billing cycles on the new logic: consistently 14 to 18. Same NiceJob subscription. Same ad spend. Same job volume. Different trigger logic and a 72-hour window.

The 48-hour phone calls on non-clean jobs didn't generate many reviews. That's not what they're for. What they generated was callbacks that didn't become complaints. A few of the red-tag jobs converted into panel replacement bookings within two weeks because the dispatcher call was the first time anyone had walked the customer through the scope without the stress of the initial discovery moment.

One thing worth naming now: the first version of this didn't work because my techs weren't filling out the status field consistently. That's a separate problem with a separate fix. But the architecture is only as good as the data going in.

---

What to Actually Do Next Monday Morning

Four things, in order.

First: Pull the last 30 job-completion messages that fired from your system. Housecall Pro, Jobber, Service Fusion — they all log this. Cross-reference against which jobs produced a review, a callback complaint, or a repeat booking within 60 days. That's the audit. Two hours. Do it before you touch a single setting. You need to know what your current system is actually producing, because "our follow-up is good" is a guess until you pull the log.

Second: Add a job status field to your close workflow. Four values, your labels, but the logic has to separate "everything went smoothly" from "something complicated happened." One field, one branch.

Third: Move your review request timing to 72 hours on clean closes. Suppress it entirely on flagged jobs until the dispatcher call happens.

Fourth: Put the 48-hour phone call on your dispatcher's task list for every non-clean close. Don't replace it with a text. A text is not the same thing — I've watched texts go unanswered on jobs where a two-minute call would have caught a problem before it became a one-star review.

Do the audit first. Let what you find tell you how fast to move on the rest.

---

FAQ

My review count is fine — 20 or 30 total on Google. Why does this matter?

Total count is a snapshot. What matters is whether it's growing. Thirty reviews accumulated over three years reads differently to a customer than 30 reviews with eight in the last 90 days. If your count is growing by one or two a month, you're losing ground to shops actively working the close. The number I care about isn't the total; it's the monthly rate.

We use Jobber, not Housecall Pro. Does the same logic apply?

Same logic, slightly different wiring. Jobber has custom job fields and connects to Zapier cleanly. You'd build the same branch — status field on close, Zapier routing clean versus flagged, suppressed review request on the flagged track. The specific field setup is a little more manual than Housecall Pro's workflow automation, but it's not an afternoon project. You need the trigger to know what happened on the job, not just that time has elapsed.

How do I get my techs to actually update the job status field?

It's a training problem, not a software problem. My techs didn't fill out the field consistently until I started showing the team monthly how many of their jobs generated repeat bookings — and the status field was the thing that made that number trackable. When techs can see their own job outcomes in a number they can influence, compliance goes up. Also make the field a required close step in the app, not optional. A dropdown with four choices takes four seconds. A paragraph prompt sits empty.

Is the 48-hour follow-up call worth it on a small ticket job under $300?

It's not about the ticket size. It's about the status field. I've had $185 troubleshooting jobs that ended with a customer confused about what we found — those are exactly the calls that turn into bad reviews if nobody follows up. If the tech flagged it as anything other than clean, it gets the call. Dollar amount doesn't change that.

We tried NiceJob and our review count didn't move. Is the tool the problem?

Probably not the tool. I use NiceJob. It does what it says it does. The timing and the trigger logic are almost always the issue. If you're sending requests at 24 hours on every job regardless of outcome, you're hitting some customers before they're ready and some at exactly the wrong moment. Move to 72 hours on clean closes only, suppress everything else, and give it a full billing cycle before you blame the platform.

At what point does the 48-hour call actually scale?

It scales when you have a dedicated dispatcher or CSR. At five trucks you should already have one. I ran the solo version myself for the first 90 days at Reeves Electric — that's where the NPS recovery came from — but it's not sustainable at volume. The handoff works if you've trained your dispatcher on what the call actually is. It's not a sales call. It's a "did everything make sense, do you have questions" call. Some of them take three minutes. Some take ten. Train on the ten-minute version; the short ones don't need training.