Ninety seconds. That is roughly what it takes a host to answer the phone, tell someone you are open until ten, hang up, and remember where they were in seating the party of six standing in front of them.
The call was maybe forty seconds. The other fifty are the part every labor calculation leaves out, and they are the reason a phone agent feels like a bigger change than the arithmetic suggests it should be.
Building the number from the bottom
There are three inputs, and only one of them is worth arguing about.
The first is call volume, which your phone provider or POS reporting can give you. Use answered calls plus missed calls, since the missed ones are going to become answered ones and they carry their own handling time. If you have never pulled this number, how many calls does your restaurant miss is where to start.
The second is handle time. Do not estimate this. Sit with a timer for two shifts and clock thirty calls from ring to hang-up. You will find a wider spread than you expected: someone asking about parking is under a minute, a straightforward pickup order lands around two to three minutes with the repeat-back, and a family ordering for six with three modifications will take five or more. Take your own average.
The third is the interruption cost, and this is the contested one. It is the time between hanging up and being fully back on the previous task. At a host stand during a rush it is substantial. At a quiet counter at two in the afternoon it is close to zero.
Doing the arithmetic
Say you handle 900 calls a month across all shifts, and your measured average handle time is two minutes and twenty seconds. That is 2,100 minutes, or 35 hours.
Now the interruption piece. If you assign an extra minute to the calls that arrive during service and half of your call volume arrives during service, that is another 450 minutes, or 7.5 hours. Total, roughly 42 hours a month.
Every number in that paragraph is yours to replace. That is the point. The method is stable and the inputs are not, and a calculation an owner can redo with their own figures survives scrutiny in a way a vendor's estimate never does.
The wage question
Multiply hours by the fully loaded rate of whoever is actually taking the calls, not your average wage. At most restaurants the phone is answered by a host, a counter person, or a manager, and those are three very different numbers. If a manager is fielding catering calls, that portion of the hours is expensive and should be priced as such.
Loaded means wage plus payroll taxes and whatever benefits load you carry. Using the base wage understates the figure by a meaningful margin and it is the kind of thing a careful owner will catch.
One caution on the manager portion. Manager time is the most expensive input in the calculation and also the most tempting to inflate, because it makes the total look better. Use the hours you can actually point to on a calendar, such as the catering calls that arrive between two and four in the afternoon, rather than a general sense that the phone eats the day.
Which hours actually leave the schedule
Here is where most of these calculations quietly overstate themselves, and it is worth being blunt about it.
Forty-two hours a month is not a person. It is not even close to a person. And it is not distributed in a way that lets you cut a shift, because it arrives as ninety-second fragments spread across every service period you run.
What those fragments buy is service quality. The host stays at the door. The counter person finishes the transaction in front of them. The manager is not walking to the office mid-rush. That is real and it is worth paying for, but it does not show up on a payroll report, and calling it a labor saving in front of an owner who then looks at their payroll report is how you lose credibility.
Some hours do leave the schedule, and they are worth identifying specifically. If you staff a dedicated phone position during peak, that position may become part-time or disappear. If you pay an answering service for overnight and off-hours coverage, that line item goes away entirely, which is the cleanest saving in the whole exercise and the one covered in switching from an answering service. And if a manager is spending forty minutes a day on catering calls, that is concentrated enough to reclaim as actual management time.
The rest is quality. Present it as quality.
The revenue side, which is usually larger
Nearly every operator who runs this calculation discovers the labor number is not the interesting one.
If you were missing calls before, and most restaurants with any peak volume were, the answered-call revenue dwarfs the hours. Twenty recovered calls a month at a thirty-dollar average ticket is six hundred dollars of revenue that previously went to whoever the caller dialed second. That comparison is worked through in the ROI calculator and it is the number that usually carries the decision.
Which means the labor calculation has a narrower job than people give it. It is not the case for the purchase. It is the answer to the specific question "what does my team stop doing," and it is most useful when it is small, precise, and clearly bounded. Against a plan starting at $250 a month, forty hours of scattered interruption time is a supporting argument, not the argument. Where the whole line item sits is covered in budgeting for voice AI monthly, and the payback math is in payback period.
What to do with the number once you have it
Write it down in three parts and keep them separate.
Raw handling minutes, from your own timed sample. Interruption minutes, labeled as an estimate with the assumption stated. Hours that convert to payroll, which will be a small subset and may be zero at a single location with no answering service and no dedicated phone position.
Then go back in ninety days and measure the same three things again. Handle time will have changed, because the calls your staff still takes are the ones the agent escalated, and those skew longer and harder. Call volume may have risen, because a phone that gets answered gets called. Both of those move the calculation, and neither of them moves it in the direction the original estimate assumed.
If after ninety days the payroll-convertible hours are still zero, that is not a failure. It means you bought service quality and recovered revenue rather than labor, and you should say so out loud rather than keep pointing at a hours-saved figure that never reached the schedule.