Team and Training

Server Performance: How to Coach the Stages of a Table

Many restaurants judge server performance on one number: total sales per shift. It is easy to pull and easy to rank, and in my experience it tells a manager almost nothing about what to teach. A table that runs long or leaves with a smaller check went wrong somewhere specific, at the greeting, the order, the pause before dessert or the wait for the check. If you only see the total, coaching turns into an argument about who got the better section. Break the table into stages and you have something concrete to work on.

I'm Richard Yu. I have more than 15 years of startup and go-to-market experience, much of it in customer success and solutions engineering, where my job was turning usage data into coaching conversations that helped customers stay and grow. Today I'm a co-founder and the Managing Partner for marketing and operations at the restaurant I co-founded in San Francisco, where I built Regular, the hospitality operating system, one service at a time. This is how I think about coaching servers stage by stage. You can find me on Instagram.

Why Sales Per Server Is the Wrong Scorecard

Sales per server mixes together things a server controls and things they do not. Section assignment, party size, the night of the week and how fast the kitchen is moving all show up in that one number. A server with a strong total might be rushing tables and leaving dessert unsold. A server with a weaker total might be doing excellent work in a section full of two-tops.

The bigger problem is what a leaderboard does to the team. When the only scorecard is a ranked list, the conversation after service becomes "I had the worst section" instead of "what happened at table 12." Servers learn to defend their number rather than improve their work, and managers end up refereeing.

I prefer to ask two questions about each table. Where did the time go, and where did the spend go? Both questions have answers you can find, and both lead to something a server can practise at the next shift. A total cannot tell you that a server is slow to greet or quick to drop the check before the guest has decided on dessert. Stages can.

The Stages of a Table From Seated to Departed

A table is a sequence of handoffs, and each one is a place where time and spend are won or lost. The exact cut points vary by concept, but this is the breakdown I use for a full-service dinner:

  • Seated to greeted: how long the guest sits before someone from the team says hello.
  • Greeted to first order: drinks, water, the first recommendation.
  • Order to first course: partly kitchen, partly how cleanly the order was fired.
  • First course to mains: pacing, clearing, the check-in.
  • Mains to dessert or check drop: the moment most often rushed or forgotten.
  • Check drop to paid: how long the guest waits to leave.
  • Paid to departed: the goodbye and the reset.

Each stage has its own coaching point. A long seated-to-greeted gap is a section awareness issue. A short mains-to-check gap with no dessert sold is a selling and reading-the-table issue. A long check-to-paid gap is often the most frustrating part of the night for a guest, and it is entirely fixable. I wrote more about the overall rhythm in Table Turn Time: Pacing Service Without Rushing Guests.

How to Spot the Stage Where Each Server Loses Time

Once you see stages, compare each server to the team, not to a perfect standard. Look at the typical time each server spends in each stage for the same party size, then look for the one stage where they sit furthest from everyone else. Usually there is one. A strong server might be fast everywhere except check drop to paid, because they disappear to the bar once dessert lands.

Spend works the same way. Look at where the check grows. One server sells a second round of drinks reliably but rarely offers dessert. Another opens strong with the wine list and then goes quiet.

At Naehas, where I built the solutions engineering and customer success functions, the useful conversations with customers rarely came from top-line numbers. They came from finding the one step in a workflow where people stalled. The floor is no different. In Regular, the Server Performance app splits each table turn into stages from seated to departed and ranks servers by pace and spend, with room for coaching notes. You can do a rougher version by hand: time a handful of tables per server over a week and write the stages down.

Running a Five-Minute Coaching Conversation After Service

Coaching works best when it is short, specific and private, and when it happens close enough to the shift that the server remembers the tables. I keep it to five minutes at the end of the night, one server at a time, rotating through the team over the week. Here is the routine I would hand a manager tonight:

  1. Start with one thing that went well, tied to a specific table and stage.
  2. Name one stage to work on, not three. "Check drop to paid ran long on your four-tops."
  3. Ask what happened before offering a theory. Often the server already knows.
  4. Agree on one behaviour to try, for example, carry the card reader back with the check.
  5. Write a one-line note so the next manager on shift can follow up.
  6. Check the same stage next week and say so out loud when it improves.

The point is that the server leaves with one thing to do differently, not a ranking. If the same stage comes up across several servers, it belongs in training rather than individual coaching. I cover that in Server Training Manual: What to Put In and What to Cut.

Which Servers Work Best Together

Individual stages only tell part of the story. Many restaurants run servers in pairs or teams, with shared sections, a server and a backserver, or a closer covering for an opener. Some combinations make both people better. Others quietly slow both down, often because two people each assume the other is handling the check-in or the water refill.

The way I look at it is simple: compare how each server performs when they work with a particular partner against how they perform alone. If a server's mains-to-dessert stage tightens up whenever they share a section with one colleague, that pairing is worth scheduling on your busiest nights. If two strong servers both slow down together, split them up and use each to anchor a newer teammate.

Regular's Staff Pairings app does exactly this comparison from shift data. Without software, you can get a rough read by keeping a notebook of who worked together each night and how the stages went. One caution: if pairings affect how tips are shared, check your local rules on tip pooling before changing how you schedule or split.

Technology That Makes Servers Better, Not Busier

I hold a firm view on this. Technology at the table should give servers more time with guests and give managers better questions to ask. It should not put another screen between a server and the person they are serving, and it should not turn a shift into constant tapping and logging.

What technology should do at this moment:

Capture the stages from systems the restaurant already uses, such as reservations, point of sale and scheduling, so servers do not have to record anything extra. Show the manager where time and spend went after service, not during it. Keep coaching notes alongside the data so follow-up does not depend on memory.

What it should not do: rank servers on a public screen mid-shift, ping them with alerts while they are at a table, or replace the conversation with a dashboard. The number is the start of the coaching conversation, not the conversation itself.

That is how I built the at-the-table apps you can see in every app in Regular. If you want to see stage-by-stage coaching on your own data, you can book a Regular demo.

Frequently Asked Questions

How do you measure server performance in a restaurant?

Rather than relying on total sales alone, break each table into stages: seated to greeted, greeted to first order, order to first course, first course to mains, mains to dessert or check, check to paid, and paid to departed. Compare each server to the team for the same party size and look for the one stage where they differ most. That gives you something specific to coach.

What should a server coaching conversation include?

Keep it to about five minutes, private and soon after service. Start with one specific thing that went well, name one stage to improve, ask the server what happened, agree on one behaviour to try at the next shift, and write a short note so another manager can follow up. Check the same stage the following week and acknowledge progress.

Why is sales per server a poor way to rank servers?

Sales totals mix in factors servers do not control, including section, party size, night of the week and kitchen speed. They also hide where a table slowed down or lost spend. Ranking on totals tends to turn coaching into an argument about sections instead of a conversation about what to do differently at a specific stage of the table.

How do you decide which servers to pair together?

Compare how each server performs when working with a specific partner against how they perform alone, looking at pace and spend by stage. Schedule pairings that lift both people on busy nights, and split up pairs that slow each other down. If pairings affect how tips are shared, check local tip pooling rules first.

Regular

Run the Night Your Regulars Remember

One suite of apps that connects your reservations, POS and schedule, from before the meal to the next visit.