Stop Outsourcing Your Phones Before You Fix Your Intake

When I launched Reeves Electric in February 2022, I thought I'd planned for everything. Branded vans, a Jobber account, a pricing model I'd spent three months building. What I had not planned for was the substance of what happened when the phone rang.

Not the logistics of answering it. The actual call — what questions to ask, what information determined whether we could help, what a good handoff to dispatch looked like. None of that was documented. I assumed twelve years in the trades meant I already knew how to do it.

My NPS in month two was a 4. A customer called at 9:30pm, irritated, because I'd promised a callback that didn't happen. Not because anyone dropped the ball on purpose. Because there was no process.

That's what I come back to every time a shop owner asks me whether to hire someone to handle their phones.

---

Why You Have to Answer the Phones Before You Can Hire Someone to Do It

The question almost always comes framed as a capacity problem. "I can't be in the truck and on the phone at the same time." True. But that's a different question than who should be answering right now.

Most owners outsource intake before they've figured out what a good intake call actually looks like. They hand a number to a virtual answering service, or they hire a part-time receptionist, and they assume the problem is solved because they're no longer the one picking up. What they've done is delegate a process that doesn't exist yet.

I had the software stack. I had the pricing model. What I didn't have was a single documented answer to: what does a good first call with a Reeves Electric customer look like? Not in my head — written down, repeatable, trainable.

I knew how to run electrical. Running intake is a different skill.

---

What's Actually Breaking on That Call

After the month-two NPS disaster, I started answering every call myself and keeping notes. Not formal documentation — a running list. What the customer called about, what I asked, what I wished I'd asked, whether it booked, and if it didn't, why not.

Then I got CallRail set up and started listening back.

What I heard was not what I expected. Hold times that felt short to me were clearly long to customers. My scope questions were vague — "so what's going on with your panel?" — instead of specific enough to tell me whether we could actually help. When customers asked for a price I couldn't responsibly give over the phone, I hedged in ways that read as evasive. And almost nobody on my end was confirming appointment details before hanging up, which meant we showed up to wrong addresses and jobs where the customer thought it was a free estimate when we'd said it wasn't.

None of this was obvious until I heard it back.

Once a week now I sit down with my dispatcher and we listen to four calls — two that booked, two that didn't. We tag what went well and what went wrong. That practice, more than any marketing change I made that year, moved our booking rate. Not because calls got dramatically better overnight. Because we could finally see what was happening on them.

Most small-shop revenue leakage happens in the first call. Most owners have never listened back to a single recording.

The owner who tells me their booking rate is "around 60%" is estimating. The owner who can pull up last month's numbers — booking rate, call volume, the two biggest reasons jobs didn't close — is making decisions from something real. Those are different businesses.

---

The Real Cost of Getting This Wrong

Here's the math I use when I think about the intake handoff.

My panel upgrade plus EV charger install is a $4,800 to $9,200 job depending on the load calc and how cooperative Austin Energy is on the meter base. That's not a generic number — that's what I quote in East Austin on a 1970s ranch with 100-amp service and the garage on the wrong side of the house. Lose a handful of those a month to a broken intake process and the number gets big fast. The expense of a virtual answering service shows up on the P&L. The lost jobs don't.

Here's what I hear from shops with bad virtual receptionist results: "we tried a service and it didn't work, booking rate was terrible." When I ask what script they gave the service, the answer is usually some version of "they took a name, number, and the problem." That's a message pad. Not a script.

The service didn't fail. They executed exactly what they were handed.

---

The 90 Days: What You're Actually Building

By month nine of running Reeves Electric, NPS was 81. Not because I hired better people. Because ninety days of answering my own calls gave me data on what a good intake interaction actually required at this shop, for these job types, in this market.

For an EV charger call, the specific output was five questions I now ask before anyone gets in a truck:

1. How old is your panel, and do you know the brand?
2. What's your current service size — 100 amp or 200 amp?
3. Where's your garage relative to your meter base?
4. Do you know if your panel has open breaker slots?
5. Have you pulled permits for electrical work at this house before?

Those five questions tell me, before a single windshield mile, whether this is a $900 job or a $6,500 job. They exist because I answered enough calls to learn which information gaps were costing us money. You can't download that question set from a template. You build it by listening to what breaks.

---

Turning Notes Into Something Trainable

Ninety days of call recordings and a failure log are raw material. Here's how I turned them into something I could actually hand to a hire.

The intake framework starts with the diagnostic questions — the five to seven things on your most common job types that actually change how you'd quote the job. Not "what's the problem." Specific questions where the answer shifts the scope. Write them down in order.

Then write out what your intake person is not allowed to improvise. For us: quoting a price before the screen is complete, promising a same-day arrival before dispatch confirms it, and telling a customer the panel "probably just needs a breaker replaced" before a tech has seen it. Those are the three places improvisation has cost us reviews. Write yours down. They're in your failure log.

Then build a confirmation loop at the end of every call. Customer name, address read back, appointment window confirmed, what they should expect next. Every call.

Once those pieces exist, you have something trainable.

On sequencing: hire your bookkeeper before your CSR. I didn't do it that way — I hired both in the same week in 2022 and neither job got done well for the first month. Two new people learning two new workflows while I was trying to train both of them was a bad idea. Bookkeeper first. Their workflow is discrete and their mistakes don't hit the customer. CSR second, once the intake script is documented and tested.

When you bring on a CSR, set up a dedicated CallRail number for their calls. Record everything. Review it the same way — weekly, four calls, two booked two not. The monitoring doesn't go away because you hired someone. It becomes the feedback loop that makes the hire work.

I tried CallRail's Premium tier AI call scoring for a quarter. For a five-truck residential shop it wasn't worth the upgrade. The recordings are the product. Listen to them yourself.

---

What to Do Monday Morning

First: Get CallRail set up, base tier, this week. One tracking number on one channel — wherever you think most calls come from. Don't analyze yet. Just get the recordings rolling.

Second: Build a failure log. Plain spreadsheet, five columns: date, call source, what broke, did it book, notes. Fill it out for every call this week and look at it Friday. Not a month from now. Friday. That cadence — weekly numbers, weekly review — is how you stop being surprised by predictable problems.

Third: Friday, listen back to two calls that didn't book. Write down the one question you wish you'd asked that would have changed how that call went. That question goes in the intake framework you're building.

Ninety days of that and you have real data. The script that makes your eventual hire trainable is in that data.

---

FAQ

My virtual receptionist has been handling calls for six months and booking rate is low — is it the service or my intake?

Probably the intake, but you can't know without the recordings. Pull the last 30 calls and listen to 10. If booking failures cluster around vague scope questions, missed callbacks, or CSRs volunteering prices they shouldn't — that's a script problem. If calls are dropping, holds are long, or customers sound confused about basic info — that's execution. Those are different fixes. The shops I've talked to with this complaint usually find the script is the culprit. The service executed exactly what they were given.

How long does it actually take to build a working intake script from scratch?

In my shop, about six weeks of answering calls and logging failures was enough to see the real patterns. Another two weeks to write the framework down and test it out loud. Call it 60 days to a first draft worth handing to someone. Ours wasn't perfect at 60 days. It was documented, which meant when something broke after we hired, we were fixing a known problem instead of diagnosing a mystery.

What if I physically can't answer every call for 90 days — I'm a one-truck shop and I'm in the field?

Answer the calls you can. Let the rest go to voicemail and listen back to both — the ones you took and the ones you missed. Missed calls are data too. You'll see fast which hours you're losing the most leads. If you're running one truck you're probably not fielding 40 inbound calls a week, so the volume is workable. The point isn't to answer every call perfectly. The point is to hear enough calls to know what good looks like before you hand it off.

At what truck count does it make sense to hire an in-house CSR vs. keep a virtual service?

I made the move at four trucks. But it's less about truck count and more about inbound call volume and average ticket. Once you're taking enough complex calls that your booking rate depends on someone understanding your specific jobs — your service area, your pricing structure, your dispatch reality — a virtual service starts costing you more than it saves. The problem was never the model. It was handing either one a script that wasn't built for your actual work.

How do I know if my intake is broken if I don't have a baseline booking rate?

Start measuring this week. Inbound calls that end in a booked appointment divided by total inbound calls. That's your number. Once you have it, record the calls and listen to them. You can't fix what you can't see, and you can't see it without the recording.

Should my CSR be quoting prices on the first call?

On clearly scoped, consistently priced jobs — a ceiling fan swap, a GFCI replacement — a range is fine. Anything involving a panel, a service upgrade, new circuits through finished walls — no. My rule: if the price can move more than 30 percent based on what's behind the wall, the tech quotes it. Write that line explicitly into the script. Improvised pricing from an undertrained CSR is where bad reviews come from, and I've got the call recordings to prove it.