Key takeaways
- Export at least four weeks of call records and count unique callers in each hour, or each 15 minutes around peaks, for every day of the week.
- Multiply calls per hour by average call length to get phone minutes per hour; that is the workload you staff for.
- When phone minutes approach half an hour in any hour, the host cannot also run the stand; plan a backup or overflow route.
- Restaurant demand is seasonal, so recheck the curve before your busy season and holidays[1][2].
- Restaurant staff average part-time hours, so the phone plan must work with whoever is on that shift[3].
Why hour-by-hour volume matters
Most restaurants staff the host stand for covers and walk-ins, then hope the phone fits around them. But call volume has its own curve, and it does not always match the dining room. Booking calls can bunch up before lunch, pickup orders can surge an hour before the dinner peak, and calls about tonight's tables arrive while the stand is busiest. If you know the curve, you can put a second pair of hands on the phone for the ninety minutes that need it instead of overstaffing the whole shift.
The broad shape is predictable from how people eat: in federal survey data, lunch and dinner were the most common fast food occasions, at 43.7% and 42.0% of adult eating occasions[4]. Your own phone line will have its own version of that pattern, and the only way to see it is to count.
Where the data comes from
| Source | What you get | Good for |
|---|---|---|
| Cloud VoIP portal | Every call with time, caller number, duration, outcome; often an hourly report | Most restaurants; easiest export |
| Traditional carrier | Call detail records on request or an itemized bill | Basic counts; may need cleaning |
| POS-linked phone | Call log next to orders and guest history | Comparing calls with orders by hour |
| Call tracking numbers | Calls by source and time | Marketing attribution plus volume |
| Manual tally sheet | Marks by hour and call type from hosts | Call reasons, or when no system data exists |
System data tells you when calls arrive and how long they last. A tally sheet tells you why. Use both for at least a week to understand the mix: a peak that is mostly pickup orders calls for a different fix from one that is mostly table requests.
Build the curve in six steps
Export four weeks
Four weeks gives each weekday four samples, enough to see a stable pattern. Skip holiday weeks for the baseline.
Clean the data
Remove internal and staff calls, known supplier numbers and obvious spam. Merge repeat attempts from the same number within fifteen minutes.
Bucket by hour
Count unique callers for each hour of each day. Around your peaks, use 15-minute buckets; an hour can hide a sharp 20-minute spike.
Average by weekday
Average the four Mondays, the four Tuesdays and so on. Note the busiest single hour each day.
Add call length
Calculate the average duration of answered calls for each hour. Order calls are usually longer than hours questions.
Make a heat map
In a spreadsheet, days across the top and hours down the side, with conditional formatting from light to dark. The dark cells are where the phone needs help.
Turn calls into phone minutes
Call counts alone can mislead. Twenty quick hours questions take less time than eight pickup orders. Multiply calls per hour by the average call length in that hour to get phone minutes per hour, which is the workload your staff actually carry.
- Example: 12 calls in an hour at an average of 1.5 minutes is 18 phone minutes. A host can absorb that around greeting and seating.
- Example: 18 calls at 2.5 minutes is 45 phone minutes. One person would spend most of the hour on the phone, so the stand needs a backup or overflow route.
- Example: 30 calls at 3 minutes is 90 phone minutes, more than one person can physically answer. Calls will ring out unless two people or a service share them.
A practical rule of thumb: when phone minutes approach thirty in any hour on a busy shift, the host is no longer able to do both jobs well. Treat that as the trigger for a named backup, a second ring group or overflow answering during that hour.
Read the shapes on the heat map
Every restaurant's map is different, but a few shapes come up often, and each points to a different kind of fix.
- A dark band before opening: guests calling to book or order ahead. Answer from the office, or make sure the voicemail offers a booking link and a callback time.
- A spike an hour before the dinner peak: pickup orders for early family meals. Put someone on the pickup line and add a menu option that routes straight to them.
- Dark cells during the peak itself: mostly calls about tonight's tables and running late. A named backup on the floor usually fixes this better than more phone lines.
- A long tail after close: next-day bookings and questions. This is an after-hours routing decision, not a staffing one.
- One unusually dark day: often a local event, a promotion or a listing that shows the wrong hours. Check what changed before you staff for it.
Match staffing to the curve
Once you have the heat map, write a phone plan for each day: who answers in each hour, who backs them up and where overflow goes. Restaurant staff average about 25 hours a week[3], so the plan must name the role, such as second host or floor manager, as well as tonight's person on the schedule.
| Phone minutes per hour | Plan |
|---|---|
| Under 15 | Host answers; no change |
| 15 to 30 | Host answers; a named backup picks up on the fourth ring |
| 30 to 50 | Second person on the phone, or overflow routing for that hour |
| Over 50 | Two people on phones, or a service sharing the load, plus online ordering prompts |
Shift start and end times are cheap levers. Moving a host's start time thirty minutes earlier to cover a pre-lunch booking surge often fixes more missed calls than adding a whole shift.
Put calls next to covers and orders
The most useful view puts three lines on one chart for each day: calls per hour, covers seated per hour and pickup orders per hour. Most POS systems export covers and orders by hour, and the call data comes from your phone report.
When the call line rises before covers and orders do, guests are planning ahead and the phone is your booking channel. When calls peak at the same time as covers, the stand is overloaded and missed calls are likely. When phone orders fall while online orders rise, guests may have given up on getting through. Looking at the three together turns a staffing hunch into evidence you can share with the team.
Seasons and holidays change the curve
Demand moves through the year. In 2025, US food services and drinking places sales were lowest in February, at $75.8 billion, and highest in May, at $92.8 billion, not seasonally adjusted[1]. Employment follows: in 2026 restaurant and bar employment peaked at 12.64 million in June, about 637,000 more than in January[2]. Phone volume usually rises with sales, especially before booking-heavy occasions.
Source: US Census Bureau, 2025[1]
Show the data
| Item | Value |
|---|---|
| February | 75.8 billion USD |
| May | 92.8 billion USD |
Single dates can swamp the line. Ahead of Valentine's Day 2026, 39% of shoppers planned an evening out[5], and the calls for that night start weeks in advance. Pull a separate curve for the two weeks before each big holiday and compare it with your baseline. Next year's phone plan for those weeks should come from that data, not memory.
Example: one week's heat map
Example: a fictional bistro exports four weeks of VoIP records. After cleaning, the heat map shows three dark zones: weekdays from 11 to 11:45 am, when booking calls arrive before the lunch host has finished setup; Fridays and Saturdays from 5 to 6 pm, mostly pickup orders; and Saturdays from 7 to 8 pm, mostly calls about tonight's tables.
In this example, the Friday 5 to 6 pm hour averages 22 calls at 2.8 minutes, about 62 phone minutes, so the manager adds a pickup-counter person from 4:45 and puts a press-one-for-pickup option on the phone menu. The weekday late-morning surge averages 9 calls at 2 minutes, about 18 phone minutes, so the fix is simply moving the lunch host's start to 10:45. The Saturday 7 pm calls go to a named floor manager on the fourth ring.
Review the curve regularly
Re-pull the data every quarter and before your busy season, and after any change that could move calls: new online ordering, a new phone menu, longer hours or a marketing push. Compare the new heat map with the old one cell by cell. If a change worked, the dark cells shrink. For the related measure of calls you are losing, see measuring missed calls; for seeing which marketing drives calls, see call tracking numbers for restaurants.
Frequently asked questions
How do I find out how many calls my restaurant gets per hour?
Export call records from your VoIP portal or carrier, remove internal calls and spam, merge repeat attempts and count unique callers in each hour for each day, averaged over at least four weeks.
When does a restaurant need a second person on the phone?
Multiply calls per hour by average call length. When that approaches thirty phone minutes in an hour on a busy shift, the host cannot also run the stand well, so add a backup or overflow route.
Do restaurant call volumes change by season?
Yes. Sales and staffing rise through spring and early summer and around big holidays, and phone volume usually follows. Recheck your hourly curve before each busy season.
References
US Census Bureau (2025). Monthly Retail Trade: Estimates of Monthly Retail and Food Services Sales by Kind of Business (1992-present) (opens in a new tab) ↑
US Bureau of Labor Statistics (2026). Current Employment Statistics: series CEU7072200001 (opens in a new tab) ↑
US Bureau of Labor Statistics (2026). Current Employment Statistics: series CES7072200002 (opens in a new tab) ↑
CDC National Center for Health Statistics (2016). Fast Food Consumption Among Adults in the United States, 2013–2016 (NCHS Data Brief No. 322) (opens in a new tab) ↑
National Retail Federation (2026). Valentine’s Day Spending Expected to Reach New Records (2026 survey with Prosper Insights & Analytics) (opens in a new tab) ↑