Your NPS Is a Phone Problem, Not a Field Problem

My NPS in month two of running Reeves Electric was a 4.

Not a 40. A 4. I had branded vans, a clean website, a pricing model I'd spent three months building, and journeymen who knew what they were doing. The work was good. The reviews were not. I spent two weeks convinced something was wrong in the field — a tech being short with customers, maybe, or the cleanup on job closeout. I rode along on jobs. Everything looked fine.

The problem was on the phone. And I had no idea what was happening there because I wasn't listening.

---

The NPS You're Reading Is Already Six Weeks Old

By the time the survey comes back, the experience that created the score is ancient history. The customer who gave you a 2 last Tuesday — the intake call that set them up to feel ignored happened six weeks ago. You've run 40 more jobs since then, all with the same intake process.

My month-two scores weren't about electricians who were rude on-site. They were about customers who called us, got a price range that didn't match reality, never received a confirmation text, showed up to the job not knowing the tech's name, and then got a change order they didn't expect. The field work was clean. The path from call to invoice was broken, and the field and intake were two different worlds as far as my shop was concerned.

They were not two different worlds to the customer.

So I took every call myself for ninety days. Dispatch, intake, the 9:30pm callback from the woman whose breaker kept tripping. I kept a legal pad next to my laptop and wrote down every moment where I felt the conversation either connect or go flat. By week six I had a legal pad and a half of notes. By week twelve I had a completely different intake process. NPS by month nine was 81. The electricians hadn't changed. The intake had.

---

The First 90 Seconds Are the Job

When a customer calls a residential service shop, they're not calling to buy electrical work. They're calling because something in their house scared them, or broke, or they've got a new car sitting in the driveway and they don't know what the install actually involves. The first 90 seconds is the moment they decide whether they trust you or whether they're still shopping.

Most shops blow this and don't know they're blowing it.

EV charger calls are where I see it most clearly. Customer calls, Tesla on the way, wants a Level 2 charger in the garage. CSR gives a range — "usually runs $600 to $1,200 depending on the panel." That range is accurate for maybe 40% of the homes we walk into. The other 60% have 100-amp services, outdated panels, a meter base on the wrong side of the house. The actual scope is a panel upgrade, maybe a service upgrade, parts and labor north of $5,000.

But the customer heard $1,200 on the phone.

When the tech hands them a revised quote on the driveway, they don't think "my CSR gave me incomplete information." They think bait-and-switch. The tech did clean, competent work. The review goes up that night anyway — and it's about trust, not the wiring.

I started recording every call in early 2022, using CallRail once I landed on the right stack. Every week, my dispatcher and I sit down and pull four calls — two that booked, two that didn't. We tag the moment in each call where the customer either relaxed or went flat. Running that practice through 2023 doubled our booking rate. Not the truck wraps. Not the Google spend. Listening to the calls.

The recording is a diagnostic, not an HR tool. You're looking for the exact second the customer stopped trusting you — and that second is almost always in the first 90 seconds.

---

Most Shop Owners Have Never Actually Heard Their Own Intake

Bad NPS is almost never a field problem. It's an intake problem that owners misdiagnose because they're not recording calls. So they fix the wrong thing — nicer uniforms, a job-complete checklist, $8,000 on a truck wrap. The score doesn't move.

In my shop, and in conversations with other owners at our revenue level, the revenue that disappears quietly disappears on the first call. Nobody knows because nobody listens back. The intake is invisible. The field is visible. Owners optimize what they can see.

Here's the Google LSA version of this, because it's specific and worth naming. If you're running Local Services Ads — and you should be, the leads are real — Google records every call that comes through the platform. You can pull those recordings today, for free, inside your LSA dashboard. I've yet to meet an owner who uses them for anything beyond disputing charges on calls that weren't real leads.

But if you're paying $180 a call on LSA (not an exaggeration in Austin right now), and your intake is fumbling 30% of those calls, you're not running a marketing problem. You're running an intake problem you're funding with your marketing budget.

---

What I Actually Rebuilt

The ninety-day stretch where I was on every call taught me one specific thing I hadn't expected. Customers weren't angry about price. They weren't angry about wait time. Call after call, the complaint underneath the complaint was the same: nobody told me what was happening next.

No confirmation. No name. No timeline. They called, talked to someone, weren't sure if they were booked or just on a list. Then a tech showed up without warning, or didn't show up, and there'd been nothing in between. That's a process failure. Not a people failure.

So I built two structural fixes.

First: every booked job triggers a confirmation text within five minutes of the call closing. Tech's first name, a headshot, the arrival window, a one-line summary of what we're coming to do. Four pieces of information. The "I don't know what's happening" feeling disappears.

Second: for complex jobs — EV charger installs, anything that might touch panel or service scope — I built a pre-quote phone screen. Five questions, about 20 minutes: panel age, current amperage, meter base location, distance from panel to garage, any known history of tripping or overcurrent issues. If the answers suggest scope beyond a simple charger run, we book a $99 site assessment, credited against the install if they move forward. The customer walks into the quote already knowing price can vary. Change orders don't feel like ambushes when you've talked about the possibility first.

The $99 credit model took me three months to get comfortable with. I thought customers would push back. Very few do. What they want is confidence that you know what you're doing before you start cutting into their walls.

---

The Software Side

The tool list for a 5-truck residential shop isn't long. The intake layer specifically is straightforward.

CallRail is the first spend. I'm at roughly $90 a month right now. Tracking numbers on every lead source — LSA, organic, Yelp, truck wraps, whatever you're running — and every call gets recorded and transcribed. The value is the recording. Not the AI scoring. I ran the Premium tier with automated call scoring for a quarter, dropped back to base because the scoring wasn't telling me anything I couldn't get from four calls on a Friday afternoon. The AI call scoring is useless for my purposes. The core product does the core job.

Before I landed on CallRail I tried routing everything through a Google Voice number and reviewing calls manually. It lasted about six weeks. No attribution, no transcripts, no channel breakdown. Don't bother.

Ops layer: Jobber or Service Fusion at the center, Zapier for the connections, QuickBooks Online for the books. Anything beyond that is overhead you don't need yet at five trucks.

For review management I use NiceJob — Podium works too, pick one. The configuration that matters is the trigger: review request fires automatically off job close, not off a dispatcher remembering to send a text. In my shop, firing within two hours of close gets meaningfully better open rates than end-of-day batch sends — that's what I see in my NiceJob dashboard, not something a rep told me. When intake gets better, close rate goes up, completed jobs go up, review requests go up. That's the sequence.

The whole stack runs me about $450 a month. At $1.4M in 2024 revenue across five trucks, I don't think about that line item.

---

Where to Start

Pull your LSA dashboard today and find four call recordings — two that converted, two that didn't. Don't change anything yet. Don't rewrite your intake script. Read the panel before you start swapping wires.

Record everything for 30 days. Get a month of calls before you restructure anything, so you're reacting to a pattern instead of a single bad interaction.

Then block 90 minutes. Pull those four calls. For each one, write down the first moment the customer either relaxed or went flat. One sentence per call. What you'll probably find: it's the same moment across all four calls. That's where you start. Though I'll say — I've talked to owners who listened back and couldn't immediately see the pattern, and they needed a second set of ears. If that's you, have your dispatcher sit in on the review. Two people hearing the same call often catch different things.

After two Fridays of that, you'll know more about your intake than you currently think you do. That's when you rebuild.

---

FAQ

I already have good reviews — does my intake process really matter if customers are happy with the work?

The reviews don't show you what you're missing. If your intake is fumbling 20% of calls — bad ranges, no follow-up, leaving people confused — those people aren't leaving bad reviews. They called someone else. Your NPS looks fine. Your booked-to-called ratio is quietly underperforming. You'd know this if you were tracking call volume against booked jobs by source, but most shops aren't running that number.

What's the fastest way to know if my intake is the problem versus something actually going wrong in the field?

Read your negative reviews word by word. Field problems show up as "took too long," "left a mess," "didn't fix it right." Intake problems show up as "never called back," "didn't know who was coming," "price was different than quoted," "felt like I had to chase them down." If your negatives are full of the second set, the field is probably fine.

Is CallRail worth it if I'm only running two or three trucks and doing under $500K?

Yes, and the argument gets stronger at smaller scale. At two trucks you have less margin for wasted ad spend. Knowing which lead sources are actually converting — not guessing, knowing — is worth more than the $50 to $90 a month the base tier costs. And if you're personally taking some of the calls yourself, you'll hear things in the recordings that are uncomfortable and useful. That discomfort is the product.

How do I get my dispatcher or CSR to take call quality seriously?

The Friday call review I run with my dispatcher is not a performance review. We're looking at calls together, tagging what went well and what didn't. When she's the one who catches something — "did you notice how the customer's tone changed when we gave the range without explaining why it varies?" — she owns that fix. I'm not delivering a verdict. We're both doing the diagnosis. That's a different dynamic than a training memo, and it produces a different result.

What does a good intake call actually sound like — is there a script?

A checklist for information, not a script for words. Scripts make your CSR sound like a call center. The checklist covers: what's happening at the house, how urgent it is, basic job type, panel age if relevant, how they heard about you, and what the customer needs to know about next steps before the call ends. That last part is what most shops skip. Ending the call with a clear next step the customer can repeat back eliminates most of the "I didn't know what was happening" complaints.

Should I be asking for the review during the call, or letting the software handle it?

Let the software handle it. Asking verbally during the call puts the customer on the spot before they've seen the finished work. The right trigger is job close — after the invoice is paid and the tech has marked the job complete in your field software. That's when sentiment is highest. In my NiceJob setup, firing within two hours of close outperforms end-of-day batch every time I've looked at the numbers. The rest is just doing the work well and following through on what you told them you'd do.