How to Measure Empty Drink Dwell Time and Refill Speed

Empty drink dwell time is the number of minutes a guest sits with an empty or near-empty glass before a server offers or delivers the next round. You measure it by marking when a glass reaches empty and when the refill or new drink lands, then averaging that gap across tables and shifts. Doing this reliably means capturing the empty-glass moment for most tables, not just the few a manager happens to walk past — which is why most operators end up reading it off their existing dining-room cameras rather than a clipboard.

Empty drink dwell time is one of the few service metrics that maps almost directly to check average and guest satisfaction. A guest with a full glass orders another. A guest staring at melting ice for ten minutes either flags someone down or just decides they're done. The trouble is that this moment — glass empty, no drink coming — is invisible in your POS until the next ring, which may never happen. This post covers what the metric actually is, how to define the start and stop points so your numbers mean something, what ranges to aim for, and the practical ways to capture it without parking a manager at every table.

What empty drink dwell time actually measures

Dwell time is the gap between two events: the moment a guest's glass goes empty (or hits the bottom-third "about to be empty" zone) and the moment the next drink arrives or a server actively offers one. It is not the same as time between rounds, which includes the time a guest spends still drinking. Dwell only counts the dead window — empty glass, no action.

Refill speed is the closely related operational number: once a refill or new round is rung, how long until it reaches the table. Together they tell you whether your slow second rounds are a floor-awareness problem (nobody noticed the empty glass) or a bar/runner problem (someone noticed, but the drink crawled out). Those are two different fixes, and you can't tell them apart without splitting the timeline.

Define your start and stop points first

Before you measure anything, get the whole team agreeing on what counts. Ambiguity here is what makes most in-house tracking worthless. A workable definition for full-service and bar settings:

• Start the empty-dwell clock when a glass reaches the bottom third or sits visibly empty — pick one and hold it constant.
• Stop the empty-dwell clock at the first server touch: an offer of another round, clearing the empty, or the next drink landing.
• Start refill speed at the POS ring (or verbal order) for the next drink; stop it when that drink reaches the table.
• Exclude tables that are clearly winding down — check dropped, cards out — so you're not penalizing good closing service.
• Treat water and non-alcoholic refills separately from alcohol if you care about beverage revenue specifically.

Write these down. The most common reason two managers report different dwell numbers for the same shift is that one started the clock at "empty" and the other at "ice only."

Rules of thumb to aim for

There's no universal benchmark, and anyone quoting a precise industry figure is guessing. But these ranges hold up as practical targets in most full-service and bar environments. Treat them as starting points to calibrate against your own concept and price point, not gospel.

The familiar "check back within two bites or two minutes" service rule is a reasonable mental model for the acknowledgment side. The point isn't to hit a magic number — it's to know your real average and watch it move when you change staffing, section sizes, or bar setup.

Three ways to capture it on the floor

Manual spot checks are where most operators start: a manager picks a few tables, watches glasses, and notes the gaps. It's free and it's better than nothing, but it samples only a small slice of tables, skews toward the section nearest the host stand, and quietly improves the moment staff realize they're being watched. Your Saturday-at-8pm number — the one that actually matters — is the one you're least able to stand around and collect.

Secret-shopper or guest-survey programs catch the egregious misses but lag by days and give you anecdotes, not a continuous average you can trend. They're useful as a backstop, not a measurement system.

The third option is reading it off the cameras already mounted over your dining room and bar. A vision model can watch for the empty-glass state and the next server interaction on every table in frame, log the gap, and roll it up by section, server, daypart, and location. Typically 80-90% of tables are measurable on existing camera angles day one, and the empty-glass and server-touch events come straight off the video — no POS ring required to start the clock, which matters because the empty-glass moment never hits your POS at all.

Pairing dwell with refill speed to find the real cause

Once you can see both halves of the timeline, the diagnosis gets simple. Long empty-dwell but fast refill-once-ordered means your floor isn't catching empties — that's a section-size, pre-bussing, or table-touch-cadence problem. Short dwell but slow refill means your team is attentive but the bar or runner system is the bottleneck. Track them as separate columns and the fix usually points to itself.

This also surfaces the retroactive case. When a guest complains about being ignored or a bad review mentions slow service, you can pull that table's timeline — empty glass at this time, next touch that much later — instead of relitigating it from memory in a pre-shift.

The honest comparison

Measuring empty drink dwell time by hand means a manager on the floor with a stopwatch and a clipboard, watching a handful of tables and missing the rush that actually drives the number. Reading it from the cameras you already run measures every table in frame, continuously, the same way every shift — un-gameable by staff who know the slow nights from the fast ones, and rolled up across every location so you can compare sections and stores instead of trading stories. Either way, the first move is the same: agree on your start and stop points, then start collecting a real average.

FAQ