Heat Wave Week: What Happens to an HVAC Phone Desk and What the Overflow Actually Costs

Send the calls your desk cannot pick up to a line that answers every time, and do it with conditional call forwarding so your existing number, your Google Business Profile, and your call tracking all stay exactly where they are. For a shop leaving roughly 55 calls a day on the floor across a five day spike, overflow coverage that week runs somewhere between about $249 and about $3,120 depending on how you are billed. That spread has almost nothing to do with answer quality and almost everything to do with whether you are billed per call, per minute, per unique caller, or a flat monthly fee.

The billing model you picked back in February decides what the best week of your year costs you.

What actually happens to a phone desk in a heat wave week?

The demand side shows up in local reporting every summer. WWBT 12 On Your Side reported on July 3, 2026 that Gilman Heating, Cooling & Plumbing of Ashland, Virginia had responded to more than 300 service calls since that heat wave began.

Notice what that number describes: service calls performed. It does not count the calls nobody picked up. That number is invisible by design. It does not show up in your CRM, it does not show up in your dispatch board, and unless you pull the carrier record you will never see it.

Here is the part every owner already knows and no vendor writes down: your phone capacity does not scale with the weather. You can add a truck. You cannot add a trained CSR who knows your service area, your capacity, and your pricing in the 36 hours between the forecast and the first 98 degree afternoon.

How many calls does the spike actually add?

This is a model built from reader supplied inputs, not a study. Every line below is a number you should replace with your own from last July.

• Normal summer day (your input): 45 inbound calls, essentially all answered.
• Heat wave multiplier (your input): 3x call volume, so 135 inbound calls a day.
• Desk capacity (your input): two people on the phone, five handled calls per hour each across an eight hour day. That is 80 calls a day. At an average handle time of 3 minutes, five calls per hour is only 15 minutes of talk time. The other 45 minutes goes to dispatch, parts calls, ETA callbacks, and taking payment, which is why the number is five and not twenty.
• Overflow: 135 minus 80 equals 55 calls a day.
• Five day spike: 275 overflow calls.

If your desk is one person instead of two, or your multiplier is 2x instead of 3x, the arithmetic changes and the conclusion does not. Run it with your own carrier report.

Evenings and the weekend sit on top of that, and heat wave calls skew late because the house does not lose the fight until about 7 p.m. Cactus reports that a plumbing and HVAC company in Utah booked about $150,000 in confirmed revenue in four months, and that about 58 percent of those captured leads came in after hours. That is one company's result, not a category average. More on that pattern in after-hours calls in home service.

What does the overflow cost under each billing model?

Same 275 calls. Same 3 minute average handle time. Four billing structures.

Competitor prices read from each vendor's public pricing page on August 4, 2026.

Two notes on the Ruby row, because it is the one line in the table that cannot be finished with public information. Ruby does not publish a per-minute overage rate on its pricing page (ruby.com/pricing, read 2026-08-04); the page directs larger accounts to contact the company. The $1,725 for 500 minutes on that same page works out to an in-plan rate of $3.45 per minute, which is arithmetic from Ruby's own published figure, not a stated overage rate. Ask for the overage rate in writing before you sign anything.

Why the per-call and per-minute lines matter more than the dollars suggest

Look at the top and bottom rows. Per-call and per-minute pricing invoice you the most in the exact week your trucks are already full, your techs are on overtime, and your cash is going out the door for compressors and capacitors. Per unique caller and flat monthly do not move at all.

The dollars are not the whole story. The incentive is. If watching a meter run at $10 to $14 a call pushes you to cap the volume, shorten the hours, or shut the forwarding off on Thursday, you have turned off the coverage in the one week it pays for itself. That is worth knowing before you pick a structure, not during.

One more thing that is true of any per-call or per-minute plan by definition: you are billed whether or not the call books. Which means the number that matters is not cost per call, it is cost per booked job. How each model treats spam and repeat callers is a separate question, broken down in answering service billing and spam calls.

What is that overflow week actually worth?

Run your own inputs from last July. Every percentage below is a reader supplied assumption, not a benchmark:

• 275 overflow calls
• x 60 percent that are genuine new service requests (your input; the rest are ETA checks, out-of-area price shoppers, vendors, spam) = 165
• x 55 percent booking rate when somebody answers and can name a real slot (your input) = about 91 jobs
• x $475 average completed repair invoice (your input) = about $43,000

Against that, the highest fully priced line in the table costs $3,117.50 for the week. Plug in your own booking rate and your own average ticket and the ratio may look different, but the question was never whether to cover the overflow. It is which billing structure still has its interests aligned with yours on Thursday of the worst week.

How do I turn on overflow coverage this week without touching my main number?

Conditional forwarding. Your main number keeps ringing your desk first. Only the calls your team does not pick up roll to the backup line.

There are two forwarding types and the distinction is the whole thing:

• Unconditional forwarding (72to set, @@PROTECT1@@73 to cancel on most traditional carriers) sends every call away immediately. Do not use this. Your CSRs stop getting the phone.
• Conditional forwarding sends a call onward only on busy or no answer. On traditional lines this is typically 90to forward on busy (@@PROTECT4@@91 to cancel) and 92to forward on no answer (@@PROTECT7@@93 to cancel). These vertical service codes vary by carrier and sometimes have to be provisioned on the account first, so confirm yours before you rely on them.

If you are on a hosted VoIP system (RingCentral, Nextiva, Ooma, Vonage, 8x8, Dialpad), you do not use star codes at all. You set it in the admin portal under call handling on the ring group or call queue: an overflow or no-answer destination, plus the ring timer.

What to set, in order

1. Point the overflow destination at the AI line.
2. Set the no-answer timer to about 15 to 18 seconds. A North American ring cycle runs about six seconds (two seconds on, four off), so that is roughly three rings. Six rings is 30 seconds, and a homeowner sitting in an 88 degree living room is already dialing the next company by then.
3. Turn on forwarding for busy as well as no answer. During a spike both lines are lit constantly, and busy is the condition that actually fires.
4. Confirm the original caller ID passes through so the AI can call back on the real number.
5. Test it yourself from a cell phone before the forecast lands, not during.

Nothing here touches your listed number. Your Google Business Profile stays as is. Your Local Services Ads and Google Ads call assets keep pointing at the same line, which matters more than most owners expect: see missed calls and Local Services Ads ranking. No carrier approval, no downtime, and you can pull it back off with a star code or one portal toggle when the wave breaks.

Why not just port the number instead?

Because a port is scheduled work, not a setting. It requires a signed Letter of Authorization and a Customer Service Record that matches your carrier's file exactly, the losing carrier controls the cutover date, and there is a live cutover window during which calls can drop. None of that belongs in the days between a forecast and a heat wave.

What should the overflow line actually do with a heat wave call?

Answering is table stakes. During a spike the sequence is what matters:

• Triage before booking: no cooling with an elderly occupant or an infant in the house goes ahead of no cooling, which goes ahead of a maintenance or quote call. The mechanics of that are covered in how AI answering triages emergency calls.
• Book against real capacity, not a fictional slot. A confirmed Thursday beats a promised Tuesday you cannot keep, because the promised Tuesday becomes a refund conversation.
• Handle Spanish. In a heat wave your bilingual CSR is not available at 8 p.m.
• Warm transfer only for genuine emergencies. Transferring everything defeats the point of the overflow line.

Cactus answers 24/7 in English and Spanish, qualifies the caller, and books the job natively into Jobber or Housecall Pro, creating the job during the call rather than dropping a message for someone to key in the next morning. That mechanism is covered in AI receptionist booking into Jobber and Housecall Pro. Pricing is a flat $999 per month with no contract, and Cactus states setup runs 48 to 72 hours. More on how the coverage works in AI receptionist for contractors.

FAQ

How much does overflow phone coverage cost during a heat wave week? For a shop leaving about 275 calls unanswered over five days, published rates put the week between roughly $249 (per unique caller, inside tier) and roughly $3,120 (per call, entry tier). A flat monthly fee adds nothing for the spike week. A per-minute plan cannot be fully priced from public information because the largest published plan covers 500 minutes and no overage rate is published.

Do I have to change my phone number to use an AI answering service? No. Conditional forwarding on busy or no answer routes only your unanswered calls to the backup line while your published number stays with your current carrier. Porting is the alternative, and since the losing carrier controls the cutover date it is the wrong move mid-season.

Which billing model is worst for a spike week? Per call and per minute both scale linearly with volume, so the invoice peaks in the same week your capacity does. Per unique caller scales more slowly because repeat callers from the same number count once in the month. A flat fee does not move at all.

Will an AI receptionist book jobs I cannot actually service that week? Only if you let it. Give it your real availability and it books into open capacity rather than inventing slots. During a spike the higher-value job is triage and an honest next-available time, not an overbooked Tuesday you have to cancel.

How fast can I have overflow coverage running before a forecasted heat wave? Forwarding itself is a star code or a portal setting on your end. The gating item is configuring the agent with your service area, pricing rules, and CRM. Cactus states setup runs 48 to 72 hours, so the practical answer is to start when you see the forecast, not when the phones light up.

Next step

If a heat wave is on your ten day forecast, get the overflow line configured before it lands. Book a demo at go.oncactus.com/book-a-demo and bring last July's call log, so we can price the actual overflow rather than a hypothetical one.

Sources: WWBT 12 On Your Side, July 3, 2026, Smith.ai pricing, Ruby pricing, Goodcall pricing. All four pages retrieved August 4, 2026. Vendor rates are subject to change without notice.