The Business Behind Home Services.

Topics  ·  About  ·  Submit a Tip  ·  Advertise

AI on Your Phones: The Seven Things It Should Never Promise

Before an AI answers a single customer call, decide what it’s forbidden to say, because a broken promise costs you far more than a missed call.

By Josh Osborne

·

· 8 min read

A homeowner talks on the phone in a kitchen while a service technician's work van waits in the driveway outside.

Before an AI answers a single customer call, decide what it’s forbidden to say, because a broken promise costs you far more than a missed call.

It’s 9:30 at night. Water is coming through a homeowner’s ceiling. They’re panicking, trying to protect their floors, and calling every plumber they can find.

Now imagine our AI answers on the first ring and says, “Don’t worry, we’ll have somebody there in 30 minutes.” Sounds great. Except every truck is already committed. That homeowner stops calling around because they believe we’re coming. Thirty minutes turns into an hour, then two, and by the time they realize nobody is showing up, they’re furious.

That’s a scenario, not a call from one of my shops. But it’s exactly why I’d rather tell that homeowner the truth and lose the job than lie to them and earn a terrible reputation. We’d have gone from a company that missed a call to a company that broke a promise in the middle of somebody’s emergency.

Like this? Get it in your inbox every week.

What changed in home services this week, and what it means for your shop.

No spam. Unsubscribe anytime.

My bias, up front

I own home service companies, and I sell lead generation. I make money when phones ring, and I lose sleep when they ring and nobody handles the call well. Weigh everything below with that in mind.

Automating a workflow is not the same as handing over the keys

I’ve spent years working with CRM systems, call tracking, automated follow-ups, marketing technology and sales processes. I’m a huge believer in automation. Getting the phone to ring is only half the battle. What happens after somebody calls decides whether you make money or just spend it.

I think AI has a massive opportunity in home services: after-hours calls, capturing customer information, scheduling follow-ups, making sure opportunities don’t fall through the cracks. But automating a workflow and giving software permission to make decisions for your company are two different things.

Think of it as a clipboard versus the keys to the truck. AI can carry the clipboard. It should never get the keys, meaning dispatch authority, pricing authority and the power to make promises you have to keep.

So before an AI answers one customer call, write the AI Never List. Here’s mine:

  1. Never promise a technician’s arrival time unless it’s confirmed against actual dispatch availability.
  2. Never quote a final repair price for a problem nobody has diagnosed.
  3. Never guarantee a repair will fix something without a qualified technician evaluating it.
  4. Never tell a customer a dangerous electrical, gas or plumbing situation is safe when the AI has no way to verify that.
  5. Never invent a warranty, discount, refund or company policy.
  6. Never pretend to be a human employee.
  7. Never get into an argument with an angry customer.

My rule is simple. AI can gather information. AI can help coordinate. AI should never make a promise the business can’t keep.

The escalate list changes by trade

If somebody is upset because their basement flooded, their technician didn’t show, or they think they were overcharged, I want a real person involved. AI should make us more responsive, not make customers fight a computer to get help. Treating every call the same is asking for trouble, so here’s how I’d split it.

Plumbing: immediate escalation for sewage backing up, a suspected gas leak, significant active flooding, or a loss of water service involving a vulnerable household.

Electrical: a much tighter leash. Sparking, smoke, a burning smell, exposed wiring or an overheating panel, and the AI should recognize the danger and direct the caller to appropriate emergency help. It should never try to diagnose electrical safety over the phone.

Roofing: escalate active storm damage, structural concerns, major leaks, and any homeowner being pressured into an immediate insurance decision. I don’t want a bot interpreting somebody’s coverage or promising what their carrier will pay.

On the other side, here’s what I’d happily automate: appointment confirmations, data entry, reminders and after-hours intake. If an employee spends three hours a day on that work and technology can handle much of it, why wouldn’t we use it?

Two more I’d add. Estimate recovery: contractors are sitting on thousands of dollars in estimates nobody followed up on, and AI can find them, send follow-ups, capture replies and alert a salesperson when someone is ready to talk. And missed-call recovery: if somebody called and nobody answered, they get an immediate acknowledgment and a path to a person. But recovering an opportunity is different from closing a complex job without a human. AI can open the door. I’d still want qualified people walking through it.

Tell them it’s a bot

I’d make it obvious in the first sentence:

“Thanks for calling [company name]. I’m the company’s AI assistant. I can help get your service request started, or I can connect you with someone on our team.”

That’s it. I don’t understand why companies are so obsessed with making AI sound human that they try to hide what it is. I don’t want a caller spending five minutes thinking they’re talking to a “Sarah” who doesn’t exist. If they ask for a human, there has to be a reasonable way to reach one.

I couldn’t care less if the customer thinks our AI sounds realistic. I care whether they trust us enough to let us into their home.

What the vendor demo won’t show you

Most owners I talk to aren’t buying AI because they’ve figured out a better customer experience. They’re buying it because somebody told them they could fire two CSRs and save $100,000 a year. Some AI vendors are feeding that fantasy. They show a demo where a bot answers three easy questions, books an appointment and sounds human, then call it a full-time employee who never sleeps, never calls in sick and costs a fraction of the money.

That’s bullshit. A great CSR knows which tech can handle a difficult job. They spot the customer who’s about to cancel. They know when to make an exception, when to pull in a manager, and when somebody calling for the third time needs more than another scripted answer. That person can save a customer relationship worth tens of thousands of dollars over the years.

So I wouldn’t open with a vendor by asking how many employees the software replaces. I’d ask what happens when it doesn’t know the answer. What happens when a customer asks for a price it doesn’t have? When every tech is booked? When someone is furious and demands a manager? When the caller says something completely unexpected? Show me those calls. A polished demo where the bot books a simple appointment tells me almost nothing.

And don’t tell me your AI handled 100 calls successfully if it made five promises the company couldn’t keep. Tell me what those five cost. The customer doesn’t care whether the mistake came from artificial intelligence or somebody behind a desk. Your name is on the truck, on the invoice and on the one-star review. You can outsource the conversation. You can’t outsource the accountability.

The math on a missed call

Most contractors dramatically underestimate what a missed call costs. They see one call. I see an opportunity we already paid to generate and may never get back. In plumbing that call might be a $300 repair, a $1,500 service job or a $10,000 replacement. In electrical, an outlet or a panel upgrade. Roofing and concrete tickets run larger still.

Use your own numbers. As an example: you spend $300 to generate a qualified plumbing lead, 70% become booked jobs, and the average completed job is $1,500. That lead is worth $1,050 in expected revenue before the cost of doing the work. Multiply by 20 missed opportunities and you have about $21,000 in expected revenue exposure, assuming none of those callers ever calls back.

That’s why arguing over whether the software costs $300 or $500 a month misses the point. The real question is how many additional profitable jobs it helps you book.

The other side

The strongest argument against all this caution is simple. A CSR who isn’t there at 6 PM costs you more than an AI that’s right, say, 80% of the time. A caller who gets voicemail dials the next company on the list. I agree with a good part of that. I’d rather have an imperfect AI capture a customer’s information at 10 PM than send that caller straight to voicemail.

But capturing a call and handling a customer are two completely different things, and contractors had better know the difference. The 80% that goes right doesn’t hurt you. The 20% that goes wrong is where the promised arrival time, the made-up warranty and the argument with an angry homeowner live. The Never List keeps the win without taking on the damage.

Who gains, who loses

Owners with real after-hours call volume gain the most, provided they set limits first. Your best customer service people gain too, if automation takes the data entry off their plate so they can spend their time on conversations that need judgment, empathy and experience.

Companies that let a bot speak freely lose, and so do the customers stuck talking to it. Will some companies eventually need fewer people doing administrative work? Absolutely possible, and I’m not going to pretend technology won’t change staffing. But replacing a $25-an-hour employee isn’t a win if the automation costs you $50,000 in lost customer relationships. I don’t want the cheapest possible customer service department. I want the most profitable customer experience we can deliver, and those aren’t always the same thing. Fire your best people before you know whether the technology performs, and you’re not innovating. You’re gambling with your customers.

Contractor action checklist: this week

  • Write your own Never List on one page. Start with mine and cut or add to fit your trades.
  • Write the always-escalate list next to it, by trade, and decide who actually picks up when a call escalates, nights and weekends included.
  • Check that any arrival time the AI gives is tied to real dispatch availability. If it isn’t, turn that feature off.
  • Write your disclosure line and your path to a human, then call your own line and test both.
  • Calculate your missed-call cost with your real lead cost, booking rate and average ticket.
  • Pick your scoreboard before launch: booked calls, missed opportunities, complaints, escalation rates and collected revenue. Payroll saved is not on the list.

What would change my mind

I’ll change my mind when the numbers prove me wrong. I don’t mean a dashboard saying the software handled 2,000 calls. Handled how? Did it book the appointment? Did the customer stay on the schedule? Did the tech run the call, did we sell the job, did we collect the money, and did the customer have a good enough experience to call us again?

  • A 90-day side-by-side of AI and human-assisted call handling at meaningful volume, showing booking rates, cancellation rates, complaints, escalation failures, completed jobs, collected revenue and gross profit.
  • AI producing the same or better customer outcomes, with fewer missed opportunities and a lower operating cost. At that point I’d expand its responsibilities. I’m not emotionally attached to a human answering every call. I’m attached to building profitable businesses customers trust.

One thing wouldn’t change even then. My best human employees don’t have unlimited authority to promise discounts, dispatch techs or guarantee outcomes, so why the hell would I give software more authority than I give my own people?

AI doesn’t run a home service company. People still have to show up, diagnose the problem, do the work and stand behind it. This week, write the Never List before you let a bot near a caller.

Josh Osborne

Josh Osborne owns six home service companies and writes Digital Insider from the owner’s chair: what’s changing in marketing, AI and private equity, and what it means for the people running the trucks.

The business behind home services, in your inbox

Your weekly read on the marketing, AI, private equity and tech changing how homeowners hire and how shops get bought.

No spam. Unsubscribe anytime.

Latest

More from Digital Insider

Run a home service business? Get the Insider.

One email a week on the marketing, AI, private equity and tech reshaping the trades. Written by an owner, for owners.

No spam. Unsubscribe anytime.