Guest experience metrics for restaurants: what to count each week

Count seven things each week: how often the wait you quote matches the real wait, kitchen ticket times, drive-thru times if you have a lane, order errors per 100 orders, the share of calls answered, new review ratings and your reply rate, and how many guests come back. Each one already sits in your POS, waitlist, phone log or review page.

These counts record what happened to guests, and they come from systems you already pay for. This guide says how to count each one, which system holds the data, and what to watch out for. It ends with a weekly sheet you can copy.

Which guest experience numbers can you measure?

Metric	How to count it	Where the data lives
Wait-quote match	Parties seated within five minutes of the wait you quoted, divided by all waitlist parties	Waitlist or reservation system
Kitchen ticket time	Tickets over your time limit, divided by all tickets	Kitchen display system reports
Drive-thru time	Average time from order point to exit, by daypart	Drive-thru timer
Order errors	Remakes, refunds and complaints for a wrong or missing item, per 100 orders	POS void and refund reports, plus a paper log
Answered-call rate	Calls answered by a person or a phone service, divided by all incoming calls	Phone system call log
Review rating and reply rate	Average stars on new reviews, and replies divided by new reviews	Google Business Profile, Yelp and other review sites
Repeat-guest share	Known guests this week who had visited in the previous 60 days, divided by all known guests this week	Reservation guest records or a loyalty program

The five-minute margin, the ticket limit and the 60-day window are our suggestions. Pick your own, write them at the top of the sheet, and keep them the same from week to week.

How do you count wait times?

Record the quote at the host stand, then the time the party sits. If you use Toast Tables, its Bookings report lists quoted against actual wait time for each booking, along with turn time, party size and source, and you can filter it by date, party size and source and download it. Our waitlist app guide compares other waitlist tools.

Split the figure by party size. Large tables wait longest for a table to open, and a quote that works for a two-top can miss badly for a six-top.

How do you track kitchen ticket times?

A kitchen display system stamps when a ticket fires and when cooks bump it. Toast's platform guide says its Ticket details report shows the time from when the first item of an order fires to when the last item is fulfilled, and a separate report gives times by prep station. Toast marks prep station times as a limited release and available only on KDS devices, so check what your account shows.

Printed tickets carry no bump time. Without a KDS, have the expo write the time on 20 tickets during the peak and work out how many ran past your limit. Our kitchen display guide covers the screens themselves. For a lane, our drive-thru timer guide explains what each timer stage measures and gives a worksheet for timing cars by hand.

How do you count order errors?

Count each order that needed a remake, a refund or a complaint about a wrong or missing item, and divide by all orders for the week. Record the channel and the cause on a sheet by the pass: misheard, wrong modifier, missed item, wrong bag.

Voids give a partial count. Toast's Sales Summary report has a Void Summary with the void amount, order count and item count, and links to the Voided Orders report. Staff also void for reasons that have nothing to do with the guest, and a fix handed over free may leave no void at all. Our phone order accuracy guide gives an error log for phone orders, and our drive-thru order accuracy guide gives an audit for the lane.

What is your answered-call rate?

Divide answered calls by all incoming calls to the restaurant's number. Most business phone systems keep a call log. RingCentral's admin call log, for example, shows whether each call was connected, accepted or missed, and lets an admin filter, download or schedule the log by email.

Count every incoming call, redials included, so the rate matches the call log. If one guest redialing three times would skew the picture, count unique callers per hour on a separate line rather than changing this rate. Split the count into answered, voicemail and missed, and by hour. Our call volume guide walks through that count with a worksheet.

If a phone service answers some calls, keep its count apart from staff. Maple's Voice and Pro plans include call analytics, and our product page suggests counting answered calls separately from completed requests and staff handoffs. An answered call that ends with no order and no answer to the question still counts as a problem.

How should you track reviews?

Each week, list new reviews on every site you check, with the star rating, the date and whether you replied. Average the stars for the new reviews only, since the overall score barely moves in a week.

Google's help page on reading and replying to reviews says you must verify your Business Profile before you can reply. Google checks each reply against its content policies, which usually takes up to 10 minutes and can take up to 30 days. The reviewer is notified and can still change the review. Count a reply on the day you send it.

Keep the count honest. Google's Maps content policy bars businesses from offering incentives for reviews, from discouraging negative reviews and from asking only happy guests for reviews.

How do you measure return visits?

You can only count guests you can recognize: a name on a reservation or waitlist, a loyalty account, or an online order account. Toast Tables' Guests report, per the same reports article, lists every guest in the waitlist and reservation system with the number of bookings and covers, leaving out cancelled bookings and no-shows. Card-based matching misses cash guests and people who change cards, so treat the figure as a trend line for known guests.

What goes on the weekly review sheet?

Every figure below is made up. Fill in your own counts, and compare each line with last week and the same week last year.

Metric	Last week (made up)	This week (made up)
Wait-quote match	Last week: 131 of 172 parties, 76.2%	This week: 144 of 180 parties, 80.0%
Tickets over 15 minutes	Last week: 81 of 1,190 tickets, 6.8%	This week: 62 of 1,240 tickets, 5.0%
Order errors per 100 orders	Last week: 24 errors in 1,790 orders, 1.34	This week: 18 errors in 1,850 orders, 0.97
Answered-call rate	Last week: 571 of 702 calls, 81.3%	This week: 548 of 640 calls, 85.6%
New reviews, average stars	Last week: 20 reviews, 4.4 stars	This week: 23 reviews, 4.3 stars
Review reply rate	Last week: 12 of 20 replied, 60.0%	This week: 19 of 23 replied, 82.6%
Repeat-guest share	Last week: 101 of 395 known guests, 25.6%	This week: 118 of 410 known guests, 28.8%

Give each line an owner, such as the host lead for waits and the kitchen manager for tickets. In the weekly meeting, pick the one line that got worse and agree on one change to test. In this made-up week, the rating slipped while every other line improved, so the manager would read the three lowest reviews before changing anything.

Published by Maple, which sells AI phone answering and ordering to restaurants. This AI-assisted guide combines Toast, RingCentral and Google help pages with an original metric table and weekly sheet. Every figure in the sheet is made up, and the thresholds are our suggestions. The guide reports no survey or study and makes no claim about any restaurant's results.