Every operator has a list of KPIs. Sales, food cost, labor, guest satisfaction, turnover. All lagging. All telling you what already happened. Useful, but you cannot act on them until the horse has already left.
The KPIs that give you an edge are the ones that tell you what is about to happen. Where to look tomorrow. Where the leak is starting before it shows up on the P&L. Every mature multi-unit operator I know has a handful of these leading indicators they track alongside the lagging ones. This post is my short list.
Lagging versus leading, quickly
A lagging indicator measures an outcome. Sales this week. Labor as a percent of sales this month. Turnover this quarter. You measure it after the fact.
A leading indicator measures something that changes before the outcome does. Forecast accuracy this week predicts labor variance next week. Modifier attach drift this week predicts food cost variance in three to four weeks. Guest wait time trend this week predicts satisfaction scores next month.
The reason leading indicators are worth more per hour of attention: you can still do something about them. Nobody can change last week's food cost. Everybody can change next week's modifier training.
By the time a lagging KPI moves, the story is already over. Leading KPIs let you edit the story while it is still being written.
Six leading indicators worth tracking
1. Forecast accuracy
The single most valuable leading indicator I know for multi-unit restaurant operations. Measured as the percent gap between the general manager's daily sales forecast and actual sales, tracked over a rolling 30 days.
Why it matters. Forecast accuracy is the leading indicator for labor variance. A general manager who forecasts within 5 percent staffs correctly. A general manager who forecasts within 15 percent overstaffs on slow days and understaffs on busy ones. Both cost margin, one in wage and one in service quality.
Target. 90 percent accuracy over a rolling 30 days for a mature unit. 85 percent for a new unit or new general manager. Below 80 percent, coaching time.
2. Modifier attach drift
Covered in more depth in the POS Data You Are Not Reading post. Attach rate on top modifiers, week over week. When it drifts, food cost is about to drift.
Why it matters. Cheese add-on attaching on 25 percent of burgers this week versus 40 percent last quarter is either theft, comping, waste, or a menu-behavior shift. All four are worth investigating. Food cost variance follows three to six weeks later.
3. No-show rate on the schedule
Percent of scheduled shifts that get called out or no-showed, weekly, per unit.
Why it matters. Rising no-show rate is the earliest leading indicator for turnover, morale problems, and service failure. It also predicts scheduling accuracy problems, because a general manager who overstaffs by habit will not push back on marginal call-outs, which teaches staff that no-shows are consequence-free.
Target. Under 4 percent weekly. Over 6 percent, investigate. Over 8 percent, there is a leadership issue.
4. Guest wait time trend
Average wait for a table (in a seated concept) or average ticket time (in a quick-service or counter concept), weekly.
Why it matters. Wait time creep is the leading indicator for guest satisfaction. Guests do not fill out a survey saying they had a long wait. They just do not come back. Wait time creep shows up three to eight weeks before repeat visit rates decline.
5. Punch-in variance from schedule
The percent of shifts where staff punch in more than 5 minutes off the scheduled start or off the scheduled end, weekly.
Why it matters. Punch-in variance is a discipline signal. If half your staff are punching in 15 minutes early, someone is not enforcing the schedule and labor cost is bleeding. If half your staff are punching out late, you are running out of daylight on the closing side.
6. Open ticket age at peak
The average age of open tickets on the kitchen display at the peak of the meal period, weekly.
Why it matters. Rising ticket age at peak predicts guest complaints two weeks out, comps four weeks out, and turnover on the line six weeks out. It is also the cleanest signal for BOH capacity strain.
Fig. 1 · Each leading indicator predicts something two to eight weeks out.
How to install a leading indicator
Do not install six at once. Install one. Get it habitual. Then add the next.
The pattern:
- Pick one leading indicator. For most operators I start with forecast accuracy.
- Instrument it. Pull the data, calculate the number, make it visible daily.
- Put it on the dashboard. See the separate post on the general manager dashboard.
- Ask about it in the weekly one-on-one. Same trick. What was your forecast accuracy last week? What is driving the miss? What are you doing about it?
- Wait 60 days for the habit to form.
- Add the next leading indicator.
Six leading indicators, installed one at a time on a 60-day cadence, gets you to a full leading-indicator practice in about a year. Fast enough to matter. Slow enough that each one actually sticks.
What not to track
Just as important. The KPIs I have watched operators track that produced almost no operational value:
- NPS at the unit level. Volatile, easy to game, does not drive decisions weekly. Track it quarterly at the group level.
- Cost per acquisition. Relevant for marketing teams. Not a field ops KPI.
- Social media engagement. Not a leading indicator for anything a general manager can control.
- Employee satisfaction survey scores. Useful annually. Not a leading indicator. Actual behavior (no-show rate, punch-in variance, turnover) leads the survey by six months.
- Any KPI without a target. A number without a target is not a KPI. It is a metric. Metrics decorate. KPIs act.
The forecast accuracy story
One extended example. When I took over the 21-unit Hana Group region, most units had forecast accuracy around 75 percent. Labor was running 4 to 6 points above the group benchmark. Everyone thought labor was the problem. Labor was not the problem. Labor was the symptom.
Instrumented forecast accuracy across all units. Put it on the dashboard. Made it question one in every weekly one-on-one. Within 90 days, forecast accuracy was above 85 percent across the region. Within another 60 days, labor was inside the benchmark, without a single headcount change.
Nothing about labor got directly worked on. The leading indicator did the work. Every general manager, forced to think about forecast accuracy every week, started paying attention to the demand curve. When they paid attention to the demand curve, they staffed to it. Labor followed.
That is the whole game with leading indicators. You do not fix the outcome. You fix the thing upstream of the outcome. Cheaper, faster, more durable.
The point
Lagging KPIs are the P&L talking to you about what already happened. Leading KPIs are the operation talking to you about what is about to happen. Both matter. Most operators over-invest in the first and skip the second.
Pick one leading indicator. Instrument it. Ask about it. Wait 60 days. Add the next. Inside a year you have a leading-indicator practice most operators in your competitive set do not have. That is where the compounding is.