Kitchen ticket time control: traditional method vs the Masterestaurant method

Kitchen ticket time control is won by measuring station by station, never by finished plate. The traditional method clocks the full ticket, discovers the delay after the server has come back twice, and corrects it by shouting; the Masterestaurant method installs four capture points —order in, station start, pass, handoff— with a target per station and a daily numeric checkpoint, so the delay shows up while you can still do something about it. For a restaurant with an active delivery channel this is not a matter of pride: the prep time you declare on Rappi or Uber Eats feeds the courier assignment algorithm, and promising 18 minutes when your kitchen delivers in 27 costs you rating, visibility and reviews. Verdict: install station-level measurement in 14 days, govern with p90 per station instead of averages, and keep the average for the monthly report, never for the service floor.
On an ordinary Thursday at 8:40 p.m., the kitchen of a 70-seat restaurant in Chapinero had nine open tickets and no idea which one had been sitting longest. The chef swore everything left in fifteen minutes; the KDS showed 31 minutes at the ninetieth percentile and three tickets past 50. That gap between what the chef believes and what the clock records is, after twenty years in this trade, exactly where the night's margin disappears.
Kitchen ticket time control stopped being an internal matter the moment the local digital channel entered the equation. Your ticket time is no longer audited only by you: Uber Eats audits it when it decides whether to send a courier, Google audits it when a guest writes «they took forty minutes» in a review that stays pinned to your Maps profile, and the guest at table nine audits it by comparing the wait with the place next door. According to Anne McBride, vice president of programs at the James Beard Foundation, operational consistency sustains a kitchen's reputation far more than any signature dish, and the clock is the cheapest instrument there is for measuring it.
I got this wrong for years: I treated timing as a way to squeeze the team, and I sold it that way. It works in reverse. A kitchen that does not measure lives in permanent tension because nobody knows whether the night is going well until a guest complains; a kitchen that measures works calmly because it owns a number that tells it where it stands. Masterestaurant built its ticket time protocol on that idea, with four captures per order and a target per station, and what changed was never speed. Variance changed.
Side-by-side comparison
| Traditional method | Masterestaurant method | |
|---|---|---|
| Capture points per ticket | ✕1 point (pass window), logged by eye in roughly 80% of shifts | ✓4 points: order, station start, pass and handoff, timestamped to the second |
| Metric that governs service | ✕Shift average, which hides the 10% of tickets behind 70% of complaints | ✓p90 per station; target 18 min hot line, 6 min cold, 9 min grill |
| When the delay is detected | ✕When the server complains: 12 to 18 minutes after it started | ✓90 seconds past target, with a visual alert on the KDS |
| Prep time declared on delivery apps | ✕A fixed value set 14 months ago and never revisited (typically 15 min) | ✓Weekly recalibration against real p90; maximum tolerated gap of 3 min |
| Measured effect on Google reviews | ✕Between 18% and 25% of 1-2★ reviews mention waiting | ✓Wait-related mentions under 7% after 90 days of station-level measurement |
| Associated kitchen training | ✕A two-hour verbal induction, no written standard per dish | ✓Spec sheet per dish with a target time plus 4 hours of timed practice |
| Running without the owner present | ✕p90 degrades 22% on shifts without the head chef | ✓Deviation under 6%, because the standard lives on the board, not in the chef's head |
Step 1: capture four timestamps per ticket, not one
The first deliverable of this guide is a ticket that records four timestamps instead of one: arrival at the KDS, station start, plate finished, and server pickup. Measuring only the full ticket exit gives you a comforting number that hides where the time went, which is why that 70-seat kitchen in Chapinero believed it was firing in 15 minutes while the 90th percentile read 31. With four stamps you can see the ticket landed at 20:41, the grill started it at 20:49 —eight dead minutes in queue— and the plate sat four minutes in the window waiting for a server. Verification: export 200 tickets from one service and confirm all four marks exist on 95% or more. If stamps are missing, your problem is not the kitchen, it is the capture. Every station needs its own target, because cold apps and grill do not play in the same league: a salad leaves in 4 minutes and a medium-rare loin needs 18 to 22, and averaging both produces a number that describes no real guest.
Step 2: set a target per station and stop governing by the house average
Set the target from your own kitchen history, not from a figure found online: take the 50th percentile of the last two weeks per station, cut 15%, and that is your starting target. A traditional restaurant turns a table in 1.5 to 2 hours according to The Restaurant HQ, and lunch for two runs about 45 minutes, so every minute you shave in the kitchen becomes an available seat rather than an efficiency medal. The deliverable is a sign at each station showing two numbers: target and yesterday's p90. If your average is 19 minutes and your p90 is 38, you do not run a fast restaurant: you run a fast restaurant with a furious minority, and that minority opens Google Maps the moment it gets home. The 90th percentile describes what the worst tenth of your guests actually lived, and that tenth weighs more on your reputation than the other nine combined.
Step 3: govern with the 90th percentile, because that guest writes the review
According to Anne McBride, vice president of programs at the James Beard Foundation, operational consistency sustains a kitchen's reputation far more than any signature dish, and p90 is the cheap way to measure consistency. Calculate it per service and per station: sort your times low to high and take the value that leaves 90% below it. Measurable deliverable: a weekly sheet with p50 and p90 per station, plus the gap between them in minutes. Fixing a single station drops your full ticket time by 5 to 9 minutes without hiring anyone, and that is the highest return available in kitchen operations. The bottleneck almost never sits where the chef believes: usually it is the flat top sharing fire with sautés, the oven opened twelve times a service, or the station doing mise en place at midnight on a Friday. Pick the station with the widest gap between p50 and p90 —not the slowest on average— because that is where the variance lives.
Step 4: attack the bottleneck station, one only, for two weeks
Change one thing: more mise, another pan, moving a product to another station, or splitting the dish into two passes. Two weeks, one variable, measurement before and after. The deliverable is a p90 comparison of the treated station against its own baseline, expressed in minutes and in extra tables per service. The most expensive mistake is using the clock as a whip: the moment your team understands the number exists to scold them, they start marking exit before plating and you lose the data forever. I got this wrong for years, selling measurement as pressure when it is exactly the opposite. The second mistake is measuring only at peak and believing that is your restaurant; measure Tuesday at 13:10 too, because comparing valley against peak tells you whether the problem is method or staffing. The third is changing two things at once and never knowing which one worked. And the fourth, the quietest: counting time to the window instead of to the table, while the plate goes cold for four minutes waiting on a server who is running a check.
The four mistakes that ruin this measurement
Masterestaurant instruments the pickup precisely for that reason. Suppose you leave p90 at 38 minutes for a full quarter. The dine-in guest waits, never returns and never complains —roughly 96% of unhappy customers say nothing, they simply vanish—; the delivery guest does write, and your Maps listing collects three one-star reviews containing the word «slow». Platform algorithms read your ticket time and send you fewer couriers, so orders leave colder and the reviews get worse. With brands above 68% off-premise sales growing 3 percentage points faster than the rest in 2024, according to Black Box Intelligence, you are bleeding the channel that grows. None of this shows up in your P&L as a line called «times»: it shows up as flat sales and a food cost creeping up because you keep throwing away reheated product. A timing system that never translates into cash dies by the third month, so run the arithmetic in front of the team.
Step 5: turn minutes into money before the management meeting
If your six-top at dinner holds 90 minutes according to The Restaurant HQ and you cut 7 minutes of kitchen time, you gain 7.8% of capacity on that table; with 40 dinner tables and a 22 dollar average check, that is about three extra tables per service and roughly 66 dollars a night. Multiply by 26 nights and you have 1,700 dollars a month that came from a stopwatch, not a campaign. Add the waste savings: food service threw away 290 million tonnes of food in 2022 according to UNEP's Food Waste Index Report 2024, and part of that is plates returned cold. Deliverable: one slide with minutes, tables and money. You finished this guide well when you can answer five questions without opening a system. First: do 95% of last service's tickets carry all four timestamps? Second: does every station have a sign with its target and yesterday's p90, and did the team read it before doors?
Closing checklist: how to know everything landed
Third: did the gap between p50 and p90 on the station you treated drop at least 5 minutes against its baseline? Fourth: does someone review the number every day at the same hour, in three minutes, without turning it into a meeting? Fifth: can your manager state, in money, what the quarter's improvement was worth? If the fourth one fails, the system dies on its own within six weeks. Pick tomorrow the station with the widest gap and give it a target today, before dinner service. MEASURE BY STATION, NOT BY TICKET. The full-ticket average is a consolation number: it hides the fact that your grill delivers in 22 minutes while the cold line plates in 4, and that the guest gets a lukewarm salad because it waited on the steak. Split the measurement by station and you find the bottleneck is almost never where the chef assumed, and fixing one station shaves 5 to 9 minutes off the entire ticket without hiring anyone.
Four differences that change the outcome
That is marginal efficiency in its purest form. GOVERN WITH p90, NOT WITH THE AVERAGE. An average of 19 minutes alongside a p90 of 38 means you run a fast restaurant with a furious minority, and that minority writes the reviews. The ninetieth percentile mirrors the experience of your worst-served tenth, which is precisely the group that sits down to rate you on Google Business Profile. Change the number you watch and you change what you fix. TREAT PREP TIME AS A LOCAL MARKETING VARIABLE. Delivery platforms route orders toward venues whose declared time matches their fulfilled time; promise 15 and deliver in 28 and the courier waits, the order flags late, your storefront slips down the carousel. Raising your declared prep time feels like surrender and it is the opposite: an honest venue at 22 declared minutes, consistently met, outranks one promising 15 and missing 40% of the time.
Four differences that change the outcome — in practice
WRITE THE STANDARD SO IT OUTLIVES THE CHEF. A kitchen's operational maturity test is the Tuesday the head chef does not show. If p90 spikes that day, you do not own a process; you own a person. The spec sheet with a target time per dish, backed by timed kitchen training, converts individual talent into process standardization, and it also protects food safety, because exposure time in the 5 °C to 60 °C danger zone stops being improvised.
Criterion-by-criterion analysis
What the average kitchen does todayTraditional
- Clocks the whole ticket and never learns which station held it up.
- Declares a fixed prep time on Rappi and Uber Eats that nobody has reviewed in over a year.
- Mistakes the average for control: the shift 'went fine' even though eight tables waited 40 minutes.
- Fixes backlogs by raising a voice instead of reassigning a station.
- Leaves inventory control out of the analysis, so a missing prep item reads as a slow cook.
- Loses the standard the night the head chef is off, because nothing was written down.
What the Masterestaurant protocol installsMasterestaurant
- Four timestamps per ticket and a distinct target for each station.
- Station p90 on the pass board, reviewed at the close of every service.
- Digital-channel prep time recalibrated each Monday against last week's real data.
- A numeric checkpoint per step: no shift closes without logging three figures.
- Mise en place tied to inventory control, with 12 critical items counted before doors open.
- A spec sheet per dish carrying the target time, which is what keeps the floor running without the owner.
Side-by-side comparison
| Traditional method | Masterestaurant method | |
|---|---|---|
| Capture points per ticket | ✕1 point (pass window), logged by eye in roughly 80% of shifts | ✓4 points: order, station start, pass and handoff, timestamped to the second |
| Metric that governs service | ✕Shift average, which hides the 10% of tickets behind 70% of complaints | ✓p90 per station; target 18 min hot line, 6 min cold, 9 min grill |
| When the delay is detected | ✕When the server complains: 12 to 18 minutes after it started | ✓90 seconds past target, with a visual alert on the KDS |
| Prep time declared on delivery apps | ✕A fixed value set 14 months ago and never revisited (typically 15 min) | ✓Weekly recalibration against real p90; maximum tolerated gap of 3 min |
| Measured effect on Google reviews | ✕Between 18% and 25% of 1-2★ reviews mention waiting | ✓Wait-related mentions under 7% after 90 days of station-level measurement |
| Associated kitchen training | ✕A two-hour verbal induction, no written standard per dish | ✓Spec sheet per dish with a target time plus 4 hours of timed practice |
| Running without the owner present | ✕p90 degrades 22% on shifts without the head chef | ✓Deviation under 6%, because the standard lives on the board, not in the chef's head |
The figures behind this guide
“For a year and a half we told everyone we plated in fifteen minutes, and the board showed a p90 of 34. The first thing we did was raise our Rappi prep time from 15 to 24 minutes, which genuinely hurt. By week three the algorithm gave us visibility back, orders climbed 19% and one-star reviews about waiting dropped from nine a month to two. Then we pulled p90 down to 23 by splitting grill from sauté, and only then did we cut prep time to 20 with a straight face.”
How to install kitchen ticket time control in 14 days
Before touching anything you need three items: a KDS or at minimum a tablet with a clock at the pass, your 12 best-selling dishes with their sales share, and the prep time currently declared on each delivery platform. Across three full services, log four timestamps per ticket: order in, station start, pass, handoff to table or courier. DELIVERABLE: a sheet with at least 150 measured tickets and p90 calculated per station. Common mistake: sampling only Fridays, which skews the baseline. Numeric checkpoint: fewer than 150 tickets or fewer than three services and you do not move on.
With the baseline in hand, assign a target per station rather than one global number. Our starting standard is 18 minutes hot line, 9 grill, 6 cold and 4 desserts, all measured at p90 and never as averages. Then write spec sheets for those 12 dishes carrying target time, gram weight and food cost, which must land at 32% or below; when a dish misses its time because the technique forbids it, redesign the dish or move it to another station. DELIVERABLE: 12 spec sheets signed by the chef, time and cost included. Checkpoint: no sheet above 32% food cost and no target time above current p90 minus 15%.
Half the delays a manager blames on slow cooks are really stockouts: the cook stops to hunt, to portion, to thaw. Define the 12 critical items feeding your highest-volume dishes, set a minimum prepped level per service and have them counted before doors open, with a signature. Use that same count to log walk-in temperatures and product pull times, because food handling and the clock belong to one file. DELIVERABLE: an opening checklist with 12 counts and 4 temperature logs. Checkpoint: zero stockouts across the 12 critical items for five consecutive services before you reach day 10.
Now touch the digital channel. Raise the prep time declared on Rappi, Uber Eats and DiDi until it matches your real p90 plus two minutes of slack, however uncomfortable the number looks, and mirror it on your Google Business Profile if you publish pickup windows. In parallel, run four timed kitchen training sessions: the cook executes the dish with the clock in view and compares against the spec sheet. DELIVERABLE: prep time updated on all three platforms and 4 sessions with per-cook time records. Checkpoint: gap between promised and fulfilled time under 3 minutes on 85% of week-three orders.
This protocol survives or dies at closing. Every night the manager on duty logs three numbers: service p90, tickets over target, and minutes on the slowest ticket. Three figures, thirty seconds, nothing else. Every Monday, read the week's series against your declared prep time and pick a single adjustment, not five. When p90 falls steadily for three weeks, trim the declared prep time and win back the commercial promise. DELIVERABLE: a logbook with 7 closes per week and one documented adjustment. Checkpoint: 30 consecutive closes with no gaps; miss one and the protocol is not yet a habit, which means the owner is still the process.
And with AI?
Forecast demand, adjust purchasing and automate operations checklists. Diego F. Parra is an expert in AI applied to restaurants.
Free tools to apply this now
Ecosystem tools that hold the measurement together
Timing data that never reaches the cash register produces pretty dashboards and the same bad margin. These three pieces of the Masterestaurant ecosystem close the loop between the clock at the pass, the cost of the dish and the month's cash, which is where you finally learn whether the effort paid.
Frequently asked questions about kitchen ticket time control
How long should a dish take to leave the kitchen in 2026?
How long should a dish take to leave the kitchen in 2026?
In full-service dining, National Restaurant Association benchmarks place the acceptable wait for a hot entrée near 13 minutes, and the Masterestaurant standard sets 18 minutes of p90 on the hot line, 9 on grill and 6 on cold. Measure the ninetieth percentile, not the average: averages hide exactly the guests who write one-star reviews.
Is ticket time control worth it if my restaurant lives on delivery?
Is ticket time control worth it if my restaurant lives on delivery?
It matters more, because the prep time you declare on Rappi, Uber Eats or DiDi feeds the algorithm that assigns couriers and storefront position. With 28% of regional orders arriving late, a venue promising 22 minutes and meeting it outranks one promising 15 and missing; measured honesty is a visibility tactic, not a moral gesture.
How does ticket timing relate to food safety?
How does ticket timing relate to food safety?
They are one file read from two angles. The danger zone between 5 °C and 60 °C allows a maximum of 2 hours under the FDA Food Code, and a kitchen that never clocks anything cannot say how long a product sat out of refrigeration. Installing timestamps on tickets and on prep improves food handling and ticket time with a single effort.
Do I need a QR menu, and can I drop the physical menu to speed up service?
Do I need a QR menu, and can I drop the physical menu to speed up service?
You need BOTH, and dropping the printed menu speeds up nothing: it removes control of the experience. The printed card governs service rhythm, menu narrative and suggestive selling; the QR handles delivery, accessibility, price changes and analytics. At Masterestaurant the verdict is a physical menu at the table with the QR as a complement, each in its own role.
How do I know whether my kitchen has enough operational maturity?
How do I know whether my kitchen has enough operational maturity?
Look at the Tuesday your head chef is off. In a traditional kitchen p90 degrades around 22% on shifts without the chef; with a written standard and timed training the deviation falls below 6%. That gap is the real test of running without the owner, and you earn it with spec sheets and checkpoints, not with charisma.
Sector data 2026 (official sources)
Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.
| Metric | Benchmark 2026 | Source |
|---|---|---|
| Ghost kitchens que operan con plataformas de terceros | Más del 70% (global) | Market Growth Reports 2024 |
| Consumidores que consideran esencial pedir para llevar (EE. UU.) | 52% (67% millennials, 63% Gen Z), 2024 | National Restaurant Association 2024 |
| Reacción negativa a los precios dinámicos/surge en restaurantes (EE. UU.) | 64% reacción negativa; 81% cambiaría de hábito para evitarlo | National Restaurant Association (Restaurant Technology Landscape) 2024 |
| Consumidores a favor de precios dinámicos (EE. UU.) | 61% a favor (Gen Z 71%, millennials 67%), 2024 | National Restaurant Association (Restaurant Technology Landscape) 2024 |
| Rotación por hora en servicio limitado (EE. UU.) | 135% en Q3 2024 | Black Box Intelligence 2024 |
| Rotación por hora en servicio completo (EE. UU.) | 96% en Q3 2024 | Black Box Intelligence 2024 |
Related content
Grow your restaurant with the Masterestaurant method
Applied in +8.400 restaurants across 43 countries.
