Saturday brunch, 9:45 AM. The tickets are piling up. Average ticket time just hit 30 minutes. One order’s been waiting 45 minutes. Guests are getting frustrated. Your team is drowning.
You know what went wrong. The grill station got slammed. But why did it happen? And more importantly, how do you stop it from happening next Saturday?
The traditional approach (and why it fails)
Most restaurants handle this the same way: Have a post-service debrief. Everyone agrees to “do better next time.” Maybe move a few things around on the line. Hope it works.
Here’s why that doesn’t help: You’re guessing. You don’t actually know which dishes created the bottleneck, or where the mise en place was inefficient, or whether having two people on grill would have solved it.
So you commit to doing better. And next Saturday, the same thing happens.
What one Boston restaurant did instead
As mentioned on Elided Opinions, a restaurant in Fenway had exactly this problem. Their peak brunch ticket times were hitting 31 minutes consistently. The team was demoralized. Comps were piling up.
The owner did something different. After service, he uploaded three things to ChatGPT:
- Toast product mix report (what dishes sold)
- Ticket fulfillment data (how long each order took)
- Photos and videos from the line during service
His prompt: “Here’s our data and some images from today’s brunch service. WTF went wrong?”
Within 5 minutes, he had answers:
- 67% of dishes were coming from the grill station
- 25% of those grill orders were wraps
- The mise en place placement was inefficient – high-volume ingredients required longer reaches
- The station wasn’t designed for two people to work it during peak times
From diagnosis to solution in one week
They redesigned the line layout based on AI’s analysis. Split the grill into two sections – one for griddle, one for wraps. Reorganized mise en place so frequently-used items were closest. Added a steam table to keep scrambled eggs at temperature without overcooking.
The results the following Sunday:
- Average peak ticket time: 9 minutes (down from 31)
- Consistency: Every time period stayed within an 11-minute band
- No more wild swings from 7 minutes to 32 minutes depending on when you ordered
But here’s what the sous chef cared about most: “The anxiety level dropped as much or more than the average ticket time.”
Why this works (and what you’re probably missing)
You already have the data. Your POS system tracks every order, every ticket time, every item sold. You have photos in your phone from when things go sideways.
What you don’t have is time to synthesize it. To spot the pattern that 67% of orders hit one station. To realize your most-sold item has the worst placement in your mise.
That’s not a discipline problem. It’s a bandwidth problem.
AI can analyze in 5 minutes what would take you hours to piece together manually – if you even had hours, which you don’t because you’re running a restaurant.
This isn’t just about brunch
The same approach works for:
- Analyzing why Friday dinner service consistently runs longer than Saturday
- Understanding which menu items create the most tickets-per-hour load
- Identifying prep work that’s eating your team’s time without adding value
- Figuring out why one server’s section turns tables faster than everyone else’s
You take the data you already have. You upload it. You ask specific questions. You get actionable answers.
Start here
Pick your worst service. The one that still makes you wince to think about.
Pull the reports: product mix, ticket times, whatever your POS gives you. Take a few photos of your line setup.
Ask ChatGPT: “What bottlenecks do you see? Where would you make changes?”
You’ll probably be surprised by what it catches that you missed.
This isn’t about teaching you to use ChatGPT. It’s about having someone who knows how to ask the right questions, interpret the data, and turn insights into actual operational changes.