How to keep good restaurant general managers

A store with a strong GM runs itself on the owner's day off. A store between GMs runs on the owner. Replacing one means recruiting, weeks of training and a stretch of weaker results, so keeping a good GM is some of the best money an owner spends.

Most GMs do not leave over a single bad week. They leave after months of small signals that nobody acted on: longer hours, a thinner crew, a bonus that never seemed to pay, and no sign of a next step. This guide covers how to see those signals early, what it costs when you miss them, how to design a bonus a GM believes in, and how to build a bench so one resignation never becomes a crisis.

What losing a GM actually costs

Owners often treat GM turnover as a recruiting expense. The recruiting fee is the smallest part. The bigger costs are the owner's time covering the store and the weeks of weaker results while a new GM learns the building.

A simple way to estimate it:

• Replacement cost = recruiting and hiring costs + training wages + owner or area-manager cover time + lost contribution during ramp-up.
• Lost contribution = weekly net sales x change in margin during the ramp-up period x number of weeks.

The last line is the one people skip. A new GM rarely runs a store as tightly as an experienced one in their first two or three months. Hourly labor drifts up because they schedule cautiously, food cost variance widens while they learn the counts, and service slows while the crew tests them.

The two-point margin loss is an assumption, and your own number may be higher or lower. Compare with what happened the last time a store of yours changed GMs. Even at half that figure, keeping a good GM is worth a raise, a better bonus or a real conversation about their week.

Signs a GM is stretched thin

• Their own clocked hours creeping up week after week.
• Schedules posted later in the week than they used to be.
• Open shifts covered by the GM rather than filled.
• Reviews mentioning short staffing or slow service.
• Shift leads leaving, or the GM no longer mentioning who they are training.
• Weekly counts skipped or rushed, so food cost variance gets noisier.
• Shorter, flatter answers in your weekly check-in.

These are prompts for a conversation, not a verdict. Ask what's going on before changing anything. Often the cause is fixable from your side: a crew position that has been open for a month, a schedule built around one person's availability, or a district manager who calls every evening.

Watch the GM's own hours

The most reliable early signal is the GM's clocked or reported hours. A GM who works 50 hours a week by choice is normal in many formats. A GM whose hours climb from 48 to 55 to 60 over six weeks is covering something. Pull their hours alongside the store's open positions and overtime. If all three are rising together, the store is short-staffed and the GM is paying for it personally.

A bonus that works

• Two or three measures the GM controls: hourly labor, food cost variance, recent rating.
• Targets set against the store's own baseline, not the brand's best store.
• Written terms, paid on time, with the math shown.
• Nothing tied to things they can't change, like rent or a brand promotion.

A bonus only retains people if they believe they can earn it. A target based on the brand's top store tells a GM at an average store that the bonus is decoration. A target based on their own store's recent normal range tells them that a better month will be paid for.

Setting the targets

A practical approach is to use the store's own trailing 12 weeks as the baseline, then set the target slightly better than that. For example:

• Hourly labor target = trailing 12-week hourly labor percentage minus 0.5 to 1 point, with a floor that still lets the store staff for its volume.
• Food cost variance target = actual food cost minus theoretical food cost, within a band such as 1 to 1.5 points, depending on your menu and how good your theoretical data is.
• Rating target = keep the recent average rating at or above its trailing level, measured on a rolling window so one bad review doesn't erase a month.

Hourly labor here means the labor your POS or scheduling tool reports from punches. It does not include payroll taxes, benefits or salaried pay, so it will read lower than fully loaded labor on your P&L. Say which one the bonus uses in the written terms, and never mix them. For more on setting a sensible labor target, see restaurant labor cost percentage.

Paying it

Pay monthly or quarterly, not annually. A year is too long for a GM to connect the work to the reward. Show the calculation on one page: the baseline, the actual, the target, the payout. If the payout is zero, say why in plain words and say what next month looks like. A bonus that quietly fails to pay is worse than no bonus.

Workload you can take off their plate

Pay is one lever. Workload is often the bigger one. Before raising a GM's pay, check what you are asking them to do that does not need a GM.

• Reports they assemble by hand that a system could produce.
• Messages from the owner, the district manager and the brand that all ask for different things on the same day.
• Calls to cover shifts that a scheduling app could post to the crew.
• Inventory counts done alone on a Sunday night.

One clear message a day beats five scattered ones. If you have more than one store, agree internally on who talks to the GM about what, and keep it to one thing at a time. How to give your GM instructions they'll follow covers how to write that message.

Build the bench

Every store should have at least one shift lead who can run the building for a day. Have them build the schedule once a month and explain the weekly numbers once a quarter. The day a GM leaves, you'll know who's ready.

A bench also helps you keep the GM. GMs who are training a successor can take a real holiday, and a GM who can take a holiday is less likely to burn out. Make it part of the GM's bonus or review: one shift lead able to open, close and run the week's counts.

If you run several stores, keep a short list of who could step up at each, and who could move between stores for a few weeks. That list is worth more than any job ad on the day you need it. Owners with several units can find more on running the week across stores in how to manage multiple franchise locations.

A new GM's first 90 days

• Days 1–30: learn the store, the crew and the weekly numbers.
• Days 31–60: own the schedule and the weekly counts.
• Days 61–90: own hourly labor and food variance, with the bonus live.

During the first 30 days, judge the new GM on whether they understand the numbers, not on the numbers themselves. Sit with them for the weekly review and have them explain each line. In the second month, they build the schedule and you review it before it posts. By the third month, the bonus should be live on the same terms as any other GM, so they know exactly where they stand.

Step-by-step: a GM retention checklist

1. List every GM, their start date and their clocked or reported hours for the last eight weeks.
2. Flag any GM whose hours have climbed for four weeks or more, and any store with a crew position open longer than three weeks.
3. Hold a 20-minute conversation with each flagged GM. Ask what is taking their time and what would help most.
4. Rewrite each GM's bonus on two or three controllable measures, set against their store's own trailing baseline.
5. Put the terms in writing and show the calculation each time you pay.
6. Name one shift lead per store as the bench, and give them the schedule once a month.
7. Cut duplicate reports and messages so each GM hears one clear priority at a time.
8. Review this list every quarter, and after any GM change, estimate what it cost.

Common mistakes

• Benchmarking every GM against the brand's best store. It makes most bonuses unreachable.
• Mixing hourly labor and fully loaded labor in one target, so the GM can't tell which number they are being paid on.
• Tying the bonus to sales alone, which pays the GM for a brand promotion and punishes them for a road closure.
• Paying the bonus annually, or paying it without showing the math.
• Treating a stretched GM's slower replies as an attitude problem rather than a workload signal.
• Having no bench, so a resignation leaves the owner running the store.
• Raising pay without fixing the workload, then losing the GM anyway six months later.

Where Franlever fits

Franlever connects to the POS, and optionally accounting, scheduling, inventory and reviews, and sends one brief each morning. Each store gets at most one lever a day, and that lever goes to the store's GM by email or text, measured against the store's own normal range. GMs can reply DONE, LATER or WRONG and never need an account. Some mornings it says nothing needs attention, which is one fewer message for a busy GM. See how AI COO works.

General information for restaurant and franchise operators, not legal, tax, accounting or financial advice. Figures in examples are illustrations, not predictions or promises of results. Laws and platform policies change and differ by location, so check the current rules and talk to a qualified advisor about your situation.