How to Measure If Your Restaurant Training Is Actually Working
Hey Team!
Most operators train and hope. Run a new hire through the menu, shadow them for a shift or two, throw them on the floor, and cross your fingers. I did it that way for years. Not because I was lazy. Because I didn't know what else to measure.
Here's what changed my thinking. I had a server at one of my concepts who aced every verbal check I threw at her. Could name the proteins, knew the specials, described dishes confidently. Comp count started creeping up after her third week. Guests were sending back a pasta because "they didn't know it had shellfish in it." She knew the dish. She clearly wasn't saying it to the table. Training looked fine. The problem was invisible until it cost me real money.
That's the difference between training activity and training outcomes. Activity is what you did. Outcomes are what changed.
So let me walk you through what I actually track now.
Quiz scores before and after training is the obvious one, but almost nobody does it right. The mistake is only quizzing after. You need a baseline first. Before you run anyone through your menu training, give them a short quiz on the current menu. Not to fail them. To establish a starting point. Then quiz again after. If someone scores 45% before and 82% after, you have something real. You can see the lift. If someone scores 55% before and 57% after, your training didn't work. That's a signal about either the content or how you're delivering it. Most operators skip the before score and wonder why some people improve and some don't.
The second number I look at is comps per cover for trainees in their first 30 days. A comp is almost always a symptom. Wrong dish, allergy surprise, something undersold or over-promised. Pull that number for new hires and compare it to your house average. If your new hires are running twice the comp rate of your experienced staff, the training gap is showing up on your P&L. Most operators know their monthly comp total. Fewer know it broken down by who generated it.
Beverage attach rate is the one that gets ignored most. If your average server attaches a beer, wine, or cocktail to 60% of covers and your new hire is at 30%, you're losing real revenue every shift they work. This isn't complicated to track. Your POS has the data. You just have to pull it by server instead of just looking at the house average. Low attach rate usually means one of two things: the server isn't confident talking about the drink menu, or they're not being trained to make the ask at all. Either way, that's fixable. But you have to see the number first.
Time-to-competent is the one I wish I'd paid attention to 20 years ago. Define what "competent" looks like for your place. At Black Barrel Tavern we set a specific threshold: a new hire needs to hit the house average on attach rate and stay under the house average on comps for two consecutive weeks. That's competent. Some servers get there in three weeks. Some take seven. Tracking that spread tells you a lot. It tells you whether your training program is consistent. It tells you which trainers are effective. And it gives you a real number to benchmark against when you change something in your training process.
Here's the part nobody talks about. All four of these metrics have to connect back to the training content itself. If beverage attach is low across the board for new hires, the question isn't "why are my new hires bad at selling drinks?" The question is "what's in our training that should be teaching this and is it actually teaching it?" Sometimes the answer is that the training doesn't address it at all. Sometimes the training covers it but in a way that doesn't stick.
At ShiftTrained, that's exactly what I built against. The platform generates quiz content directly from your menu, so when a number like beverage attach goes sideways, you can trace it back to whether the question bank is actually testing for it. That's the loop most operations don't have closed.
Now, I want to be honest about something. These metrics work better in some operations than others. If you're running a high-volume fast casual where the ticket average is $14, comps per cover might not be your signal. Time-to-competent and quiz score improvement matter more there. If you're running a full-service room with a $75 average check, attach rate and comp count are the ones that move the revenue needle most. Know your model before you decide what to track.
One more thing. These numbers don't just tell you about your staff. They tell you about your trainers. If you have two trainers and every hire who shadows Trainer A gets to competent in four weeks and every hire who shadows Trainer B takes seven, that's information. Don't ignore it. Your best trainer is a resource you should be building around, not just scheduling around.
The goal isn't to turn your operation into a metrics dashboard. The goal is to stop flying blind. Right now, most operators know if training happened. Almost none of them know if training worked. Those are completely different questions, and only one of them actually helps you run a tighter, more profitable operation.
Start with quiz before-and-after. Pull comps by hire date. Look at attach rate by server. Define what competent means and time how long it takes to get there. That's it. Four numbers. They'll tell you more than any post-shift feeling will.
Have a great day! — Terry Psaltakis
Your AI Restaurant Guy
