From Job Done to Job Verified: Why the Check After the Job Needs an Owner
How do home service companies check quality after a job? Many wait for complaints. Why silence is not verification, and why the check needs an owner.
Short answer: As I see it, most home service companies check quality after a job by waiting. If the customer doesn't call back, the job is counted as good. That is not verification, it is silence. A practical test is whether three things have happened: the technical work has been checked against a standard, the customer has been asked directly how it went, and anything that went wrong has been fixed and traced back to its cause. Review requests belong after that check, not in place of it. I think the check after the job deserves an owner and a measurable result, the way companies treat everything before the invoice, not a favor someone does when there is time.
After the job is closed, quality is often assumed. The technician leaves, the invoice is paid, the job is marked complete, and the company moves on. Too often, nobody asks the question that matters: did the work do what we promised?
That is the difference between a job done and a job verified. Below: what the evidence says, where the idea breaks, and how a $5–20M operator can start closing the gap without buying anything.
The founder's job you shouldn't delegate
I was once talking with a younger founder who was still doing all of the delivery himself. My advice was to leave the technician job first. You have to delegate the production work, the doing.
But I told him to keep one thing: quality assurance. The customer got a promise from you. Your person came and delivered. Then you have to come back and ask the customer: how was he? That part you keep.
The logic is simple. The promise is made by one person, the work is done by another. If nobody on the promise side checks the result, the two slowly drift apart, and you find out only when the customer tells someone else instead of you.
It is easy to delegate in the opposite order: keep the doing, because you are good at it, and let the checking disappear, because it doesn't feel like work. That is how quality becomes a feeling instead of a measurement.
Why silence is weak evidence
The usual way a company learns about bad work is the callback: the customer calls because something is still wrong. Callbacks matter. But a quality system built only on callbacks sees only the problems the customer can see.
In HVAC, many problems are invisible to the customer. A system can run, cool the house and still be installed wrong.
The most comprehensive public summary we found is a 2018 U.S. Department of Energy review of research on residential HVAC installation. It summarized field studies from the mid-1990s through 2016 and reported that 70–90% of air conditioner and heat pump systems in homes had at least one performance-compromising fault, either from installation or from inadequate maintenance. When duct leakage was counted, the range rose to 90–100%.
Two cautions. These are older field studies, and the review itself says the data can't reliably measure how faults combine. And a "fault" here is a performance problem, like the wrong refrigerant charge or low airflow, not a broken system. It is not a callback rate and it is not a measure of any one company.
But it tells you something important about waiting for complaints. A peer-reviewed 2020 study by researchers at the National Renewable Energy Laboratory modeled just two of these faults, indoor airflow and refrigerant charge, across U.S. single-family detached homes. It estimated that they add about 9% to the energy use of central air conditioners and air-source heat pumps, costing homeowners about $2.5 billion a year in utility bills. That is a simulation, not a measurement, but the point stands: a fault like that often doesn't make the phone ring. It can show up as a higher bill and a customer who quietly decides you were "fine" but not worth calling again.
Silence is not the same as quality. Sometimes it is just missing information.
Nobody measures callbacks the same way
If callbacks are the main signal, you'd expect a clear benchmark. In our research we did not find a published callback-rate benchmark for residential HVAC with a disclosed sample, a time period and a definition.
What does exist tells you how messy the measurement is. In an informal 2003 web poll by ACHR News, contractors reported callback rates anywhere from 1% to 50%, with most saying 4–6%. That is an old, self-selected poll, not a benchmark. A spread that wide is hard to read as one measurement. One likely reason is that companies define the word differently: one may count any second visit; another, only free return trips.
When the definition is loose, the number says whatever the company wants. Verification has to start with a written definition, not with software.
What a verified job actually looks like
A practical way to apply this is to check every high-stakes job at three layers. This is a checklist for putting the principle to work, not an industry standard and not the only way to do it.
1. Technical proof. Before the technician leaves, the work is checked against a standard someone else can read. In HVAC that can mean recorded measurements, photos of the finished work, model and serial numbers, and start-up readings. The point is not paperwork. It is that a second person, or software, can look at the job later and say "this was done right" without having been there.
2. Customer verification. Someone from the company asks the customer directly how it went. A short call: how did the visit go, is everything working, is there anything we should come back for? In the trade this is sometimes called a happy call, but its real job is to find problems while they are still small and cheap.
3. A closed loop. Every bad answer goes to a named owner who fixes it, and every fix is traced to a cause: a skill gap, a part that keeps failing, a sales promise the work can't keep, a schedule that rushed the job. If answers are collected and nobody acts on them, the company has a survey, not a quality system.
Review requests fit after those three layers, not inside them. When the work is verified and the customer has been heard, asking for a review is natural. When the review request comes first, it is marketing dressed as quality.
Why the check needs an owner
A simple exercise shows the gap. List what your company spends money and management time on before the invoice: marketing, call handling, booking, dispatch, invoicing. Then list what happens after it. If the after-invoice list is a template email and a hope that nobody calls, that work has no owner and no measurable result.
There is also a structural reason the gap tends to persist. Inside a company, the people who would check quality are often the same people whose work is being checked, or their managers. Nobody enjoys grading their own job. So the checking can get lighter every busy season until it disappears.
Whoever does the checking, independence has to be designed in. A person who reports to the team being checked has a reason to report good news. Three rules help: the same questions for every customer, raw answers the owner can see, and no say for the technician over what gets recorded.
The strongest case against this
There are three serious objections.
Customers go where it's cheaper. I have written that myself, in Customers go where it's cheaper. If price decides who gets the next job, why pay to call every customer after the last one? This is partly right. Verification won't save a company whose prices are out of line, and it won't make a price-only buyer loyal. What it can do is keep the customers who would have stayed from leaving over a problem nobody asked about.
Fixing problems doesn't buy extra loyalty. There is a popular idea, called the service recovery paradox, that a customer whose problem you fix well ends up more loyal than one who never had a problem. A 2007 meta-analysis in the Journal of Service Research, combining many studies from service settings such as hotels and restaurants, not home services specifically, found that customers whose problem was fixed well did end up more satisfied than customers who never had a problem, but were no more likely to say they would buy again or recommend the company. That is an uncomfortable finding for anyone who sells post-service QA as a loyalty machine. I think it points in the right direction, though: preventing a failure beats recovering from it, so the main value of verification is not the apology. It is catching problems early and learning enough to stop them from repeating.
It costs money and can annoy people. Every call takes time. Customers get tired of surveys. A small company may spend more on checking than it saves. That is why every job doesn't deserve the same treatment: a drain cleaning and a full system replacement don't carry the same risk.
The economics are simple to write down, even if they are hard to measure. Verification pays when the return trips it prevents, the repeat business it keeps and the bad reviews it heads off are worth more than the time it costs. For a full system replacement, the math is more likely to work, because a return trip and a bad review cost more. For a small one-time repair, it may not. Each company has to run that math with its own numbers.
So post-service QA is not automatically valuable. It is valuable when it changes what happens next: fewer repeat failures, better training, more honest sales promises, and fewer problems that go public before you know about them.
When the thesis breaks
Here is where I'd expect this idea to fail or not pay.
Low-ticket, one-time work. If the job is small, the risk is low and the customer is unlikely to come back anyway, a heavy verification process can cost more than it protects.
Owner-run companies. Where the owner still talks to every customer, the check already happens informally. The problem starts when the company grows past the owner's voice.
Data that nobody acts on. Answers with no named owner and no follow-up turn verification into theater, which is worse than nothing because it makes people feel the problem is handled.
QA used as surveillance. If technicians experience verification only as a way to catch and punish them, they will learn to hide problems instead of reporting them. The loop has to fix systems as well as people.
Review requests disguised as QA. This matters legally, not only ethically. Google's Maps review policy says businesses may not discourage negative reviews, selectively solicit positive ones, or offer incentives for reviews. The FTC's rule on consumer reviews and testimonials, in effect since October 21, 2024, prohibits fake reviews, buying reviews conditioned on their sentiment and certain kinds of review suppression. The FTC has also said review gating (asking only satisfied customers for reviews) can still violate the FTC Act. A "happy call" that sends only happy customers to a review site is not quality assurance, and it creates real risk. This is general information, not legal advice. The clean design: the same quality questions for every customer, and one rule for review requests that applies to everyone.
What this changes for operators, investors and founders
For operators. Quality becomes something you can measure and manage, not just something you believe about your company. Reviews matter, and recent ones matter most. In a 2026 survey of 1,002 U.S. adults by BrightLocal, a review-software vendor, 97% of consumers said they read reviews for local businesses, and 74% said they only care about reviews written in the last three months. It is a self-reported online survey, but the direction is clear: your reputation is judged largely on recent months, not on the last ten years. A problem you catch on a call is a problem that doesn't become this month's review.
For investors and acquirers. When you look at a home service company, ask how it knows its work is good. Ask for the written definition of a callback, the callback data by technician and job type, and what happens after a bad answer. "Our customers love us" is an opinion. Verified jobs, a callback definition and the changes made because of them are a system. In my view that is part of the quality of revenue, because work that is never checked is a liability that hasn't shown up yet.
For founders. When you build a platform, the verification layer is one of the things that should be shared and standardized, not reinvented in every unit. In the operating platform described in our thesis, quality systems sit in the network layer and outcomes and satisfaction sit in the data layer. Verification is what connects the two. And when you delegate, delegate the doing first, but keep the checking close, and build the system so it keeps happening on your busiest day.
How to start in the next 30 days
You don't need new software to begin. You need a definition, a habit and an owner.
Week 1: Define it in writing. Write down what counts as a callback: which return visits, within how long, and whether paid return visits count. Decide which job types are high-stakes, such as installs and replacements.
Week 2: Add technical proof to the high-stakes jobs. Pick the few measurements, photos and equipment details that prove a high-stakes job was done right. Make them required before the job can be closed.
Week 3: Ask every customer the same questions. For high-stakes jobs, have someone who wasn't on the job call the customer within a few days. Three questions are enough: how did the visit go, is everything working as expected, is there anything we should come back for? Ask every customer, not only the ones you expect to be happy. Keep any review request separate and the same for everyone.
Week 4: Close the loop. Every bad answer goes to a named owner the same day. At the end of the month, look at the answers by technician, job type and salesperson. Choose one cause to fix. If you have read Customers Always First, this is the same idea from the other side: the follow-up call is part of the work, not an extra.
What is supported, what is uncertain, what to test
Supported by evidence: Most residential air conditioners and heat pumps in the field studies reviewed by DOE in 2018 had at least one performance-compromising fault. Consumers lean on recent reviews to choose local businesses (BrightLocal, 2026). Customers whose problem was fixed well reported higher satisfaction than customers who never had a problem, but were not, on average, more likely to say they would buy again or recommend the company (2007 meta-analysis of studies in hotels, restaurants and other services, not home services).
My opinion, not yet proven: that many of these faults go unnoticed by customers, so silence is weak evidence of quality, and that the check after the job deserves an owner and a measurable result rather than being a favor someone does when there is time.
Uncertain: how much verification actually reduces callbacks, warranty cost and churn in home services. In our search we found no rigorous public study that measures it for our industry. The cost per verified job and where it stops paying for itself are also open questions.
What to test next: Take one high-stakes job type. Verify half of the jobs with technical proof and a customer call, chosen by a fixed rule such as alternating jobs, not by who seems likely to complain, and leave the other half as they are now. After 90 days, compare callbacks, warranty visits, complaints and repeat business. Callbacks are rare, so with fewer than a few hundred such jobs in 90 days, run it longer. If you can, don't tell technicians which jobs will get the follow-up call, or you may measure extra care on watched jobs, not the check. If the verified half isn't measurably better, the process needs to change, or the thesis does.
Frequently asked questions
How do home service companies check quality after a job?
A common approach is to rely on callbacks and online reviews: if the customer doesn't complain, the job is assumed to be good. A stronger approach adds three steps: technical proof that the work meets a standard, a direct call asking the customer how it went, and a closed loop where every problem is fixed and traced to its cause.
What is a good callback rate for an HVAC company?
We did not find a published benchmark with a clear sample and definition, and an informal 2003 trade-press poll showed reported rates ranging from 1% to 50%. Instead of chasing an industry number, define a callback in writing, measure your own rate by job type and technician, and track whether it goes down.
Should I ask for a review on the follow-up call?
You can, but keep it separate from the quality questions and apply it the same way to every customer. Google's review policy prohibits selectively asking only happy customers for reviews and offering incentives for reviews. Check the current platform rules and get legal advice for your situation.
The job isn't over when the truck leaves
A completed job in the software is a claim. A verified job is evidence. The difference is whether you find out about a problem from your own question or from your customer's review.
The work after the job is closed is not glamorous. But it is where the promise made during the sale either becomes true or quietly stops being true. Delegate the doing. Keep the checking. And build it so it still happens on your busiest day.