Modeling the payback of after-hours call coverage

Why after-hours calls deserve a model

Most owners know they miss calls after hours. Few know how many, what kind or what those calls were worth. Without that, the decision about coverage turns into a debate about gut feel and vendor claims. A simple model, built on your own phone data and a handful of public benchmarks, settles it.

After-hours demand is not an edge case. In 2025, 81% of employed Americans worked on an average weekday, compared with 30% on an average weekend day^[3]. Many homeowners can only deal with a broken appliance or a quote request in the evening or at the weekend. Trades already recognize this: BLS notes that HVAC technicians' schedules may include evening or weekend shifts^[4]. The question is not whether those calls exist but what it costs to answer them well.

The five inputs to gather first

1. 01

   After-hours call volume

   Pull four to eight weeks of call logs from your phone system and count calls outside your staffed hours, split by evening, overnight and weekend. Include calls that rang out and calls that went to voicemail.

2. 02

   Share that are real service requests

   Listen to a sample of voicemails and recordings. Remove spam, vendors, wrong numbers and existing customers calling about something that can wait.

3. 03

   What happens today

   For real requests, estimate the share that leave a message, the share of callbacks that connect and the share of connected callbacks that book.

4. 04

   Booking rate when answered live

   Use your daytime numbers for similar calls as a starting point, and adjust for urgency: evening callers with a broken system book more often than weekend estimate shoppers.

5. 05

   Gross profit per job

   Use your own average gross profit by job type, not revenue. A maintenance visit and an emergency repair are worth very different amounts.

Callback connection is usually the weakest link. Eight in ten Americans say they generally do not answer cellphone calls from unknown numbers^[5], and response speed affects outcomes: firms that reached web leads within an hour were nearly seven times as likely to qualify them as those that waited longer^[6]. Measure your own connection rate for a week; it is often lower than owners assume.

The coverage options to compare

Price each option as a full monthly cost for the hours it covers, and be specific about which hours those are. An option that covers weeknight evenings but not Saturday mornings may miss the busiest after-hours window for estimate requests. Ask every provider for the same thing in writing: hours covered, what happens on urgent calls, what the caller leaves the call with and what your team still has to do the next morning.

What staff coverage costs

If you staff evenings with your own people, use real wage data. Receptionists in the US earned a median of $18.27 an hour in May 2025^[7], and customer service representatives earned $21.53^[1]. BLS notes that customer service representatives often need to work at busy times, including evenings, weekends and holidays^[1]. Covered nonexempt employees must receive at least time and one-half their regular pay for hours over 40 in a workweek^[2], so an evening shift added on top of a full week costs more than the base rate.

Example: a three-hour evening shift, five nights a week, at the median CSR rate comes to about $1,398 a month in base wages, or about $2,098 if every hour is paid at overtime rates. That covers fifteen hours of a week that has more than a hundred hours outside a typical office schedule, and it does not include payroll taxes, breaks, training or overnight and weekend calls.

A worked model

Example: a fictional heating and cooling company logs 80 after-hours calls a month. Listening to a sample shows that 60 are real service requests. Today, 40% of those callers leave a voicemail, half of the morning callbacks connect and 60% of connected callers book. That gives 60 times 0.4 times 0.5 times 0.6, or about 7 booked jobs a month from after-hours calls.

Example, continued: with calls answered and intake completed, the owner assumes 85% of real requests finish the call with full details and 55% of those book a visit or a request that turns into a visit. That gives 60 times 0.85 times 0.55, or about 28 jobs. The difference is about 21 jobs a month. At an average gross profit of $180 per job, the recovered work is worth about $3,780 a month in gross profit.

Reading the result honestly

A model like this is only as good as its weakest assumption, and the weakest assumption is usually the booking rate once calls are answered. Owners tend to borrow their best daytime number. Evening and weekend callers are a different mix: more urgent repairs, but also more people shopping several companies at once and more calls about work you do not do. Start with a conservative figure and let the pilot tell you the real one.

• If the option only beats break-even in the strong scenario, treat it as unproven and run a longer pilot.
• If it beats break-even even in the conservative scenario, the decision is mostly about quality and control, not money.
• Check capacity. Recovered jobs only count if you have technicians to do them in the following days without pushing out higher-value work.
• Separate one-off emergency spikes, such as a heat wave or a storm, from normal months. A model built on a spike will overstate the steady value.
• Count existing customers separately. Answering their calls protects maintenance plans and repeat work even when the call does not produce a new job that night.

Finally, write down the assumptions next to the result. When you revisit the decision before the next peak season, you will want to know which numbers were measured and which were guesses.

Costs owners forget to include

• On-call pay and callout premiums for the technicians who handle escalated calls, whichever answering option you choose.
• Review time: someone reading call summaries daily in the first weeks, then weekly.
• Set-up time: writing rules, safety scripts, service-area lists and approved answers.
• Callback time for anything the option does not finish, such as requests that need a person to confirm.
• The cost of mistakes: a wrong address, a missed escalation or a booking outside your rules.

Include benefits that are easy to miss as well: fewer morning callbacks for daytime staff, a calmer on-call technician who is not answering every routine call, and better data about what customers ask after hours.

Testing the model in the real world

1. 01

   Run a baseline month

   Count calls, messages, callbacks connected and jobs booked from after-hours calls before changing anything.

2. 02

   Pilot one option for a month

   Keep everything else the same, including marketing spend, so the comparison is fair.

3. 03

   Compare like with like

   Seasonality matters. Compare with the same weeks last year where you can, or run the pilot during a steady month.

4. 04

   Decide and set a review date

   Keep the option if it beats break-even with room to spare, and review it again before your next peak season.

Whatever option you test, set up escalation properly first; see emergency escalation rules for AI answering. And plan the follow-up for callers who did not book on the first call, covered in AI follow-up texts.