How many calls does a restaurant get? How to measure your own

To find out how many calls your restaurant gets, export two weeks of incoming calls from your phone provider. Split them into answered, voicemail and unanswered, group them into 15-minute blocks, and match them against phone orders in your POS. The worksheet below walks through each count and the sums.

This page is about counting. Once you know your busiest 15 minutes, our rush-hour phone coverage guide shows how many people that block needs and where overflow calls should go.

What does Maple's phone audit say about call volume?

Maple's restaurant phone audit publishes shares of calls and gives no count per restaurant. Its data was queried on April 28, 2026, and covers paying Maple restaurants, with a 90-day window for the day and hour charts. In that group, calls built from 10 a.m., peaked at 5 p.m. local time and faded after 9 p.m. Roughly 19% of calls arrived on a closed day, before opening or after closing.

The audit describes restaurants that use Maple. It is not a sample of all US restaurants, and its figures on why people call come from keyword matching on AI-written call summaries. At restaurants that take phone orders, 39.8% of calls showed ordering intent. Use these shares to check your own pattern, and let your own log set the numbers.

Day and its share of weekly calls in the audit	Your share (your day's calls divided by your week's calls)
Monday: 11.2%	____
Tuesday: 11.9%	____
Wednesday: 12.4%	____
Thursday: 13.9%	____
Friday: 20.1%	____
Saturday: 18.0%	____
Sunday: 12.4%	____

The published shares add up to 99.9% because each one is rounded.

Where do you find your call records?

Start with your provider's incoming-call records. Confirm which numbers, time zone, routing events and retention period the export covers, and note any missing records. Two examples from the providers' own help pages:

• Ooma Office: an admin can view all company calls and download the call log as a CSV file, filtered to incoming calls. On Ooma Office Pro, a Call Resolution column marks each call as answered by a user, sent to voicemail, disconnected by the caller or unanswered.
• Google Voice: admins can search Voice log events, which include the calling number, the number called, the date and the call length, and export the results to Sheets or a CSV file.

If your line has no web portal, ask the carrier what call records it can send you. Failing that, keep a paper tally at the host stand for two weeks, with a mark for every call and a circle for every call nobody picked up.

How do you count phone orders in the POS?

Give phone orders their own label in the POS so they show up as a separate line in your sales reports. On Toast, you can create a custom dining option, such as "Phone pickup," with takeout behavior. The Sales Summary report then shows order counts and sales for each dining option, filters by hours and days of the week, and exports as a CSV or Excel file. Toast notes that the report counts a sale when the check is first opened, so an order keyed in just before a time block ends falls in that block.

A dining-option label identifies the recorded channel, not whether the order was fulfilled. Review unpaid, voided, refunded and duplicate records before using the count as an outcome. Check-opening time can also differ from the call time.

Other POS systems use names such as order type or order source for the same idea. Train staff to pick the phone label on every phone order for the two weeks you measure, or the count will come out low.

How do you find your busiest 15 minutes?

1. Open the call log export in a spreadsheet and keep incoming calls only. Delete internal and outgoing calls.
2. Add a column that rounds each start time down to its 15-minute block: =FLOOR(B2, TIME(0,15,0)), where B2 holds the call time.
3. Add a column for the day of the week.
4. Build a pivot table with blocks as rows, days as columns and a count of calls in each cell.
5. Mark the biggest number at lunch and at dinner on each day. Those are the blocks your staffing plan has to cover.
6. Repeat the pivot with only missed calls. Compare those blocks with call duration, overlapping calls, routing and staff duties before deciding whether staffing or overflow needs to change.

Hourly totals can hide bursts. Ten calls in 15 minutes may overlap; how many staff you need also depends on how long those calls take and their other duties.

How do you fill in the worksheet?

Copy this table and fill in the last column from one week of your records. The example column is made up to show the sums. Assign each incoming attempt once to lines 2–6, in this order: staff answered, including a call transferred from AI; otherwise AI answered; otherwise voicemail only; otherwise confirmed unanswered; otherwise routing unknown. A greeting or IVR alone does not count as an AI conversation. Voicemail only excludes calls already answered by staff or AI. Routing events and repeat attempts are not unique people, and an answered call does not prove the request finished.

Measure	Where to find it	Example week (made up)	Your week
1. Incoming call attempts	Provider call log, incoming only	412	____
2. Answered by staff	Include AI calls that reached staff; count each attempt once	318	____
3. Answered by AI, with no staff answer	Call records checked against the provider's definition	0 before AI launch	____
4. Voicemail only	No prior staff or AI answer in that attempt	41	____
5. Confirmed unanswered	No staff, AI or voicemail answer	53	____
6. Routing unknown	Missing or ambiguous answer records; report this count and its share separately	0	____
7. Known attempts not answered by staff or AI	(line 4 + line 5) divided by line 1; unknown routing stays separate	94 of 412, or 22.8%	____
8. Those attempts with a possible later follow-up	For each attempt in lines 4–5, look for a later staff- or AI-answered call from the same available number that day	29	____
9. Those attempts without that recorded match	Line 4 + line 5, minus line 8; unavailable or ambiguous follow-up matches stay unknown	65; request result unknown	____
10. Tagged POS orders	POS report for your phone label over the same period	176	____
11. Tagged orders per answered call event	Line 10 divided by (line 2 + line 3); a ratio of aggregate counts	0.55 orders per answered call	____
12. Busiest 15-minute block	Pivot table	Friday 6:00 to 6:15 p.m., 14 calls	____

Lines 8 and 9 count call attempts, not people. Count each missed attempt at most once in line 8; several missed attempts may share the same later call. A shared or withheld number, a callback or another contact channel can prevent a reliable match. A later answered call is a possible follow-up, not proof that the same request was resolved. Keep attempts without a recorded match unknown, not lost customers or orders. Keep any separate count of distinct numbers apart from these event counts.

Line 11 compares two totals. It does not show the share of calls that placed or completed an order, or how much staff time orders took. To measure completion, match each included call to a confirmed order within a stated window; avoid double-counting calls or tickets and check cancellations, refunds and unpaid orders. Keep unmatched records unknown. Review calls separately to classify their purpose. For common questions, our phone scripts and voicemail greetings can answer the common ones faster. For missed calls outside your hours, a missed-call text can point callers to online ordering.

How much do calls change by season?

Two weeks in one month can mislead you about the rest of the year. The Census Bureau's monthly sales for food services and drinking places, published on FRED as series MRTSSM722USN without seasonal adjustment, show how much the whole industry moves. In 2025, sales ran lowest per day in January, about $2.83 billion a day ($87.8 billion over 31 days), and highest per day in May, about $3.42 billion a day ($106.0 billion over 31 days). May's daily figure was about 21% higher than January's.

Those are national dollar sales, and calls at your restaurant may not follow them. Use them as a reason to measure twice a year, once in a slow month and once in a busy one. Your own POS sales by month from last year tell you which months those are. Also note anything unusual in the weeks you count: a holiday, a local event, a closed day or a new delivery app all change the phone.

What should you do with the numbers?

1. If line 7 is high and the misses sit in a few peak blocks, check duration, overlap and routing, then test staffing or overflow changes for those blocks, using the coverage guide.
2. If misses spread across quiet hours, look at where the phone rings. Our restaurant phone system guide covers ring groups and handsets.
3. If many calls come while you are closed, decide what those callers should hear or be sent.
4. After any change, count two more weeks and compare them with weeks from a similar time of year, so the season does not hide the effect.

If you add an answering service, keep the same call-attempt categories, denominator and time window for the comparison. Count each AI-to-staff transfer once under staff answered, keep AI-only answers separate, and report routing unknowns. Check request completion separately from these answering categories. Maple Voice answers calls 24/7 with hours and FAQ answers, automated call transfer and call analytics, for $150 a month billed monthly or $85 a month billed annually. Pro adds phone orders into supported POS systems, and the product page lists the tests to run before you connect your live number.

Published by Maple, which sells AI phone answering for restaurants. This AI-assisted guide combines Maple's published phone audit, Ooma, Google and Toast help pages and Census Bureau sales data from FRED with an original worksheet. The example week is made up. It does not report a count at any real restaurant or promise a result.