Your Bad Reviews Start at the First Phone Call — Not the Job

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

Not 40. Four. Out of 10. I had branded vans, a pricing model I'd built from scratch, techs who knew what they were doing. Good work going out the door. And customers were leaving reviews that made it sound like I'd sent someone over to rob them.

The reflex is to look at the tech. Whose name is in the review? What happened on that job? I had that conversation twice before I realized I was asking the wrong question. Then I started answering every inbound call myself — every single one, February to May of 2022 — and within three weeks I knew exactly what was happening.

The customers weren't mad about the work. They were mad about the gap between what they thought was going to happen and what actually showed up.

---

The First 90 Seconds Cost You More Than the Callback Ever Will

By the time the review posts, you've already lost. The expectation was set wrong on the call, the job delivered something different, and the customer found the only tool they had left.

Most owners read that review and think it's a verdict on the tech, the materials, the pricing. It's not. Every time I've pulled a job file and matched it to the call recording, the failure shows up in the first ninety seconds — before a truck rolled, before a tech introduced himself, before a single screw turned.

When I was answering calls personally, I kept notes on a legal pad. After about sixty calls the pattern was obvious. Customers who pushed back on price, or called to complain before the invoice even landed, had almost all gotten the same thing on intake: a booking confirmation and nothing else.

They got a time window. They got a name. No realistic scope. No price range. No explanation of what happens if the job is more complicated than expected. No explanation of what a four-hour dispatch window actually means. So they built their own expectations. And when reality showed up — which is almost never what they imagined — they were already primed to be angry.

---

Intake Is Not Scheduling

Intake isn't "get the name, get the address, put it in the calendar." Intake is "establish what this customer believes is about to happen, and close the gap between that belief and reality before anyone touches a truck."

A caller wants a Level 2 charger. They've seen the $599 install ad from a national franchise. They have a Tesla on the way. If intake is purely a scheduling function, you get the address, confirm the slot, dispatch the job. The tech shows up and realizes it's a 1972 ranch with a 100-amp Federal Pacific panel and a meter base on the wrong side of the house. The actual scope is a panel upgrade, service upgrade, new circuit. We've quoted those jobs anywhere from $4,800 to over $9,000 depending on the load calc and how cooperative the utility is on the service upgrade.

The customer's face at that moment isn't confusion. It's betrayal. Nobody on that first call gave them a reason to expect anything different from $599.

That's not a quoting problem. The tech quoted it right. It's a failure that compounded, invisibly, from the moment the first call ended.

After those ninety days, I built a pre-quote screen that every EV charger call goes through before we book. Five questions. Panel age. Approximate service size. Where the panel sits relative to the garage. Any prior electrical work done on the house. What charger or receptacle they think they need. Takes three minutes. When the screen flags a likely upgrade, the conversation shifts — we talk scope, we talk range, we offer the $99 site assessment that credits against the install if they book. The customer knows they're not getting a $599 job. They book knowing that. The angry review doesn't happen because the mismatch got caught before anyone drove anywhere.

---

What Actually Goes Wrong on the First Call

The reason owners keep misdiagnosing this is simple: they're not listening to their calls.

I sit down with my dispatcher every Friday. We pull four calls from the week — two that converted, two that didn't. We listen. We tag where the conversation shifted. We ask one question per call: what did this customer think was going to happen when they hung up?

From my own recordings, the failures cluster in four places. Vague scope — "we'll take a look at your panel" with no explanation of what that means or what it might cost. No price range — the customer is walking in with a number from Google and nobody replaced it. Dispatch window not explained — "between 8 and noon" sounds fine until the tech shows at 11:47 and the customer has been rearranging their morning since 7. No escalation path — the customer doesn't know what happens if the job grows, so the change order feels like an ambush.

None of that happens on the job. All of it happens on the call.

The review isn't about what your tech did on the job. It's about what your intake call promised — or didn't. Ninety seconds in, you either set the customer up to be satisfied or you set them up to be disappointed. By the time the truck rolls, it's usually already decided.

Getting any of this visible requires call recording. I use CallRail — tracking numbers on every inbound channel, every call recorded automatically. The AI call scoring they pitch you on upgrade? I tried it. Skip it. The recordings are the product. You cannot fix a conversation you've never listened to.

---

Your Techs Are Not the Problem

When a review comes in with a tech's name in it, the instinct is to have a conversation with that tech. It feels like accountability. It's usually wrong.

Go back to the call recording. What you almost always find is that the tech was dispatched into a situation the customer had already misunderstood. The tech showed up, did the work, quoted the actual scope. The customer said it was double what they expected. The tech wasn't on that intake call and had no control over it.

When I rebuilt Reeves Electric's intake process, I wasn't disciplining anyone. I was reading calls I'd been on myself and finding, repeatedly, that I was the one setting expectations wrong. My early intake script was: what's the address, what time works for you, we'll be there. That's a booking confirmation. It was costing me NPS points on every call.

The other thing I'd push back on: decisions made from five or six data points. Your gut after 4,000 service calls is data. Your gut after 80 is hope. Pull the recordings. Find the actual failure point. In my shop, and from what I've seen in my own audit logs going back to 2022, it traces to intake more often than it traces to anything a tech did.

---

What the 90 Days Actually Changed

February to May 2022. Every inbound call, me, personally. By month two, NPS was a 4. By month nine, it was 81.

It wasn't one thing.

The five-question screen helped — surfacing scope before we booked, so neither the tech nor the customer walked into a surprise. Explaining the dispatch window on every single call, not assuming anyone understood what "between 8 and noon" means. Stating the escalation path out loud: "If we get on-site and find the scope is larger than what we discussed, we'll call you before we do anything additional." I started saying that on every complex call in March 2022. Change-order complaints dropped noticeably within sixty days — I didn't track a hard percentage then, but the complaint calls I was fielding personally went from several a week to maybe one.

The biggest structural shift was the $99 site assessment for anything that flagged on the five-question screen. Not a revenue line — we credit it against the install if they book. What it does is change the sequence. Instead of booking at an implied price and discovering the real scope on arrival, we're booking a scoping visit at a stated cost. The customer agreed to pay for accurate information. When the real number comes back, they're in a different posture.

The bait-and-switch feeling doesn't start at the change order. It starts the moment a customer books without a realistic price range in their head.

---

What You Can Do Monday Morning

First: Get call recording running. Set up CallRail. Assign a tracking number to every inbound channel — Google ads, LSA, truck wrap, website, Yelp, all of it. Recording has to exist before any audit is possible.

Second: Pull your last five one-star or two-star reviews and read them out loud. Not to assign blame. To find the intake failure in each one. In my experience, most trace back to a price, timeline, or scope expectation that was never established on the call. Find the exact words the customer used. Ask: did our intake call ever give them an accurate version of that thing?

Third: Block 30 minutes with your dispatcher next Friday. Pull two calls that booked, two that didn't. Listen to all four. For each one: what did this customer think was going to happen when they hung up? You'll hear it. Once you hear it, it's hard to stop hearing it.

The reviews your shop got this quarter probably don't reflect your techs. They probably reflect a handful of conversations that happened before anyone drove anywhere. That's where to look.

---

FAQ

If my reviews are mostly about price complaints, isn't that a pricing problem?

Price complaints mean the customer arrived at the job with a number in their head that didn't match reality, and nobody on the intake call gave them an accurate replacement. Your prices may be perfectly calibrated and you'll still get this if the call doesn't establish a range. "I had no idea it would be that much" is an intake failure, not a pricing failure.

I don't have a dedicated CSR. The tech takes calls from the truck. How do I fix this without hiring?

Start with a written screen — five questions the tech runs through before confirming any booking. Write them, laminate them, put them in every truck. A tech taking a call from a job site is distracted by default, so shorter and more concrete is better. Record the calls. Review weekly. That tells you whether the habit is holding.

How do I get dispatcher buy-in without it feeling like a performance review?

Don't call it a performance review. Call it a curiosity exercise. "I want us to listen to four calls together Friday — not to grade anyone, just to understand what customers are actually hearing." Once a dispatcher hears their own calls, they usually fix things without being told. The Friday session does more than any training document.

Is there a benchmark for call length or booking rate?

At Reeves Electric, a well-run intake call on a complex job runs three to five minutes. Simple jobs, two to three. A call under 90 seconds that ends in a booking almost always means scope got skipped. Booking rate varies by channel — LSA calls close lower than referral calls because the customer is still shopping. What I track is booking rate by call duration, using the call logs in CallRail matched against booked jobs in Jobber. Short calls that don't book are almost always scope or price expectation failures.

My guys hate scripts and go off them anyway. What actually works?

A word-for-word script gets abandoned. A question list doesn't. Five or six things that have to get covered before the call ends — not specific language, just the required stops. Give people latitude in how they get there. Then review recordings and name specifically what went well, not just what went wrong. People adapt to a checklist. They quit a teleprompter.

When does this become a CRM problem versus a process problem?

If the intake conversation is going well but the information isn't reaching the tech, that's a CRM problem — data isn't moving from call to job. If the conversation itself is the failure — vague scope, no price range, no timeline — that's process. Fix the conversation first. Once your intake calls are consistently capturing the right information, then look at whether your software is delivering it cleanly to the field. Fixing the software before fixing the conversation just moves the mess into a database.