Every restaurant investor eventually asks the same question: how fast can we grow? The right answer is boring and specific. As fast as the bench allows and no faster. This is not the answer investors want. It is the answer that keeps operators from destroying value in year three.
I have watched groups grow through this window from both sides. As the operator inside the group deciding whether to open the next unit. And as the operator brought in to fix a group that grew too fast. The math from the outside looking in is always the same: contribution margins compressed, ratings slipped, employee retention cratered, and by the time anyone noticed, three of the five units were losing money.
What unit quality actually is
Unit quality is a composite. Not a single number. It has four components, and any one of them being weak means the unit is not really at standard yet.
- Contribution margin at brand standard: The unit is hitting its P&L target without cutting corners on food quality or service quality to get there.
- Guest ratings: A rolling 4.5+ average across the primary review platforms, with the trend flat or rising.
- Employee retention: Voluntary turnover under 40 percent annualized on hourly, under 20 percent on salaried.
- Consistency of execution: The unit runs the same way on a Tuesday afternoon that it does on a Saturday night. The variance in ticket time, food quality, and service between quiet and peak is under 20 percent.
If a unit hits all four, it is a quality unit. If it misses one, fix that one before you consider opening another. If it misses two or more, this unit is not really operating at brand standard yet and you should not be planning growth from a base that is not solid.
Fig. 1 · The four components of unit quality.
The three tests before you open
Before you decide to open unit N+1, run three tests on the current N units.
Test one: are the existing units at quality?
Apply the four-part definition above to every current unit. If two of your three current units are at quality and one is not, you have a fix-first problem, not a grow problem.
Test two: is the bench real?
Name the general manager who will run the new unit. Not "we'll figure it out" or "we can promote someone." Name them. If you cannot, the bench is not real. If you can, is that person a legitimate promotion or a stretch you are pretending is a promotion? The tell is whether you would be comfortable with them running your best current unit tomorrow. If not, they are not ready to run your newest unit either.
Test three: is the operating system portable?
If your operating system lives in the head of one person (usually the founder or the operations executive), it is not portable. It will not survive a new unit that person is not physically present at. This is the failure mode of chef-driven restaurants that try to grow. The systems that were unspoken worked in one location. In three locations they have to be written down and taught.
The 3-point rule for existing units
Look at contribution margin across your current units. The spread from the best to the worst tells you whether the group is healthy enough to grow.
- Spread of 0 to 3 points: Healthy group. All units running the same operating system with similar outcomes. Growth is defensible.
- Spread of 3 to 6 points: Watch group. One or two units are drifting. Fix before growing.
- Spread of 6+ points: Broken group. The operating system is not really running consistently. Adding a unit will make the spread worse, not better.
This rule caught me on the third opening at Zareen's. When we started planning the fourth location, the contribution spread across the three existing units was about 4.5 points. I pushed hard to fix the weak unit before opening the fourth. It took an extra 90 days. The fourth opened into a group that was consistent. If we had opened first and fixed later, all four would have drifted.
Growth pace is bench pace. If you cannot promote a real general manager for the new unit today, do not open the new unit today.
What breaks when you grow too fast
The failure pattern is consistent. It shows up in this order:
- Month 1-3 after opening: The new unit takes a disproportionate share of the operator's attention. The existing units lose oversight.
- Month 4-6: One of the existing units starts to drift. Small things at first. Labor variance up, comps up, one bad review turning into a small pile.
- Month 6-9: The drifted unit's GM starts looking for a new job because they have not gotten leadership attention. If they leave, the drift accelerates.
- Month 9-12: The new unit finally stabilizes. The operator turns back to the group and discovers two of the older units are now the underperformers.
- Month 12-18: The group is now three healthy units and two struggling ones. What was supposed to be growth has turned into a turnaround.
I have watched this exact pattern play out five different times, across three different operators. Every one of them thought their operating system would hold. None of them did. The system holds up to the number of units your bench can lead. Not more.
The case for closing before growing
Sometimes the right answer is not to grow the group. It is to shrink it. Closing a weak unit is not a failure. It is a portfolio decision.
The math almost always works. A unit that is contributing $100K a year while consuming $180K a year in operator attention, management bandwidth, and capital is a $80K net negative. Closing it returns all of that to the healthy units. In year two, the healthy units will make more than $80K in incremental contribution from the returned focus. The group is smaller and healthier.
Investors read this correctly. A group of five units where one is dragging is a group where the operator does not know when to cut. A group of four units where all four are healthy, after a clean close, is a group with a disciplined operator. The second story raises capital more easily than the first.
The bench build is a two-year project
If you want to grow, start the bench build 24 months before you want to open. This is where most operators fail. They realize at month 20 that they need three general managers for the growth plan, none of whom exist yet, and they hire externally in a rush. External hires as GMs on new openings have a much higher failure rate than promoted internals, for reasons I have written about elsewhere.
The bench build is: identify potential GM candidates at the assistant manager or strong shift lead level, give them stretch assignments (running a shift alone, running a slow day alone, running a full weekend alone), and track their progress every 90 days. Two years of that and you have a real bench. Skip it and you are hiring GMs from your competitor at $95K each and hoping they figure out your brand.
The point
The choice between opening another unit and fixing what you have is almost never as close as it looks. If the existing units are at quality, the bench is real, and the operating system is portable, open the next one. If any of the three are missing, fix first. Nine times out of ten, the fix takes 90 to 180 days and delivers more contribution than the new opening would have.
Unit count is what you show off. Unit quality is what pays the bills. Groups that optimize for count end up in turnaround. Groups that optimize for quality end up owning the market. Pick the one you actually want.