The Second Location That Looked Certain

Why this matters

A second location is the biggest bet most shop owners ever make, and it is usually made at the moment they are most confident: the first shop is full, the phone is ringing from the next town over, and the numbers in the spreadsheet look obvious. This is a reconstruction of one that went wrong, worked backwards from the owner's own records. It did not fail because the owner was careless or the market was bad. It failed in a specific, repeatable way: the plan was built entirely from the inside, and two cheap checks that would have reshaped it were skipped. Both checks are worth knowing before you sign a lease.

The record

The owner of an eighteen-person HVAC and plumbing shop opened a second location in a fast-growing suburb 55 miles from the original. Two years later, he sat down with his bookkeeper and pulled together what had been written down at each stage. The timeline they rebuilt:

MONTH  WHAT THE RECORD SHOWS
 -4    Forecast v1: satellite breaks even month 6
 -3    Lease signed: 5 yr, showroom + bay,
       personal guarantee
 -2    Manager hired away from a competitor
  0    Open. 2 techs, 1 CSR, manager
  3    Revenue at 41% of forecast
  9    Manager resigns. Owner on site 3 days/wk
 12    Home shop callbacks up from 4% to 7%
 16    Showroom closed. 2 techs now dispatched
       from a storage unit

The reconstruction is mostly about what is NOT in that record. There is no written definition of break-even, no list of what could go wrong, no comparison to anyone else's second location, and no point at which the plan would be reconsidered. Each absence turns out to matter.

Reading the forecast line by line

The forecast was a one-page spreadsheet. Three assumptions carried it.

"We already have 140 customers in that area." This was the emotional center of the decision, and the owner repeated it to his staff, his banker and his spouse. Pulled from the customer database during the review, the 140 were every address in those zip codes the shop had ever served over eight years. Customers with any job in the 24 months before opening: 46. 46 of 140 is 33%. Two thirds of the starting base the forecast assumed was people who had not called in two years or more.

"Ramp like the home shop." The forecast grew the satellite's revenue at the rate the original shop grew in its third and fourth years. But the home shop took 26 months to break even when it opened, and the years copied into the forecast were from after it was established, with a reputation, a full maintenance-agreement base and an owner on every call. The forecast assumed the satellite would start where the original shop had been after years three and four, not where it had started.

"Break-even in month 6." With no written definition, nobody could later say whether it meant covering the satellite's direct costs (techs, CSR, trucks), or those plus the lease and manager, or those plus a share of the home office's dispatch and accounting time. The three definitions differ a lot. At month 6, the satellite covered roughly the first and none of the others.

Each assumption, taken alone, was the kind of thing a reasonable owner believes. Together, they described the best plausible version of the project and called it the plan. That is the inside view, Daniel Kahneman's term for forecasting a project from its own details and the planner's own story about it, which reliably runs optimistic. The bias behind it, at the scale of a project, has its own card; see related: The Planning Fallacy in Shop Projects.

The number nobody asked for

The outside view asks a different question: how did projects like this one turn out for other people? It starts from a base rate, the typical result for the whole class, before your project's details get a vote. How to build one is covered in depth elsewhere; see related: Base Rates: The Number You Should Start From.

During the review, the owner did what he had not done before opening. He called owners he knew through his trade association and supplier counter who had opened a second location in the past ten years. He found five. Their results:

• 2 broke even (on direct costs plus their own lease and manager) within 18 months.
• 1 took about three years.
• 2 closed their second location, one at month 14 and one at month 30.
• 0 of 5 broke even in 6 months.

Five is a small sample, and it is local and anecdotal. It is still far better than none, and it says something the spreadsheet could not: in the owner's own region and trade, the typical second location took somewhere between a year and a half and three years to stand on its own, and two in five did not make it at all. Add his own first shop's 26 months, and the outside view puts planning-grade break-even at around two years, not six months.

That difference is not a detail. A plan to fund six months of losses and a plan to fund twenty-four months of losses are different plans, with different leases, different staffing and different amounts of cash held back.

The meeting that did not happen

The second missing check was a premortem: before committing, the people involved imagine the project has already failed and write down why (see related: Running a Premortem Before a Big Commitment). The owner decided alone, after a dinner with his brother-in-law, who loved the idea.

Going back through the record, the bookkeeper listed the failures that actually happened and asked, for each one, whether anyone in the shop could have named it beforehand:

• The manager leaves. Yes. He had been hired away from a competitor where he had stayed fourteen months, and the shop's office manager had raised it at the time, in passing, in the hallway.
• The owner gets pulled out of the home shop. Yes. The owner was the home shop's only estimator for replacement work. Any month spent at the satellite was a month without replacement sales at home.
• Home shop quality slips. Yes. The lead tech who moved to the satellite had been the home shop's go-to for difficult callbacks, which is the most direct explanation for the callback rate going from 4% to 7%.
• The customer base is thinner than assumed. Yes. A five-minute database query would have shown it.
• Revenue ramps slower than planned. Yes. The owner's own first shop had done exactly that.

Five of five were nameable in advance from information inside the shop. None of them required predicting the future. They required asking the people who already knew.

What it cost

The bookkeeper measured the damage in one unit: months of the home shop's average net profit before the expansion, so the cost could be weighed against something the owner understood.

• The forecast had planned for losses equal to about 1.5 months of home-shop net profit before the satellite broke even.
• Actual satellite losses through month 16, when the showroom closed: about 7 months of home-shop net profit.
• On top of that, the home shop's own profit fell during the owner's absence and the callback spike, by about 4 months' worth over the same 16 months.

7 plus 4 is about 11 months of the home shop's net profit, against a plan of 1.5, so the actual cost ran a little over seven times the forecast (11 divided by 1.5 is 7.3).

The five-year lease outlasted the showroom. The owner negotiated a sublease of the showroom space at month 20, and kept paying the difference between the lease and the sublease rent. Because he had signed a personal guarantee, that remaining obligation was his, not just the company's. How a personal guarantee works depends on the lease's wording and the state whose law governs it, and that is a question for your own attorney before signing, not after. This is not legal advice.

What the owner got right

It would be easy to read this as "second locations are a mistake." The record does not say that. The suburb really was growing. The satellite's two techs, dispatched from a storage unit and sold by the home office, were covering their direct costs plus the storage and truck costs by month 24 and are now a profitable service area. The demand was real.

What failed was the shape and the timing: a showroom, a manager and a five-year lease, committed before the business had shown it could find customers there. The outside view would not have said "don't." It would have said "fund two years, not six months," and a premortem would very likely have said "start with two techs and a storage unit, then add a showroom when the route is full," which is roughly what the owner ended up with, eleven months of profit later.

Judging a decision by its outcome is its own trap, and part of this reconstruction was separating the two (see related: Outcome Bias: Judging a Decision by How It Turned Out). The test the owner applied was: given what was knowable at month -4, was the decision process sound? On the market question, yes. On the forecast, the staffing and the commitment size, no, and all three failures were visible in documents and conversations he already had access to.

What the owner does now

For any commitment over a size he has set in advance, he runs three things, and writes each down:

• A base rate first. At least three comparable projects from people outside the shop, before looking at his own spreadsheet.
• A premortem with the office manager and the lead techs, in writing, before anything is signed.
• The smallest version that tests the idea. The question is never "should we open a second location?" but "what is the cheapest way to find out whether we can sell work there?" Two techs and a storage unit would have answered it.

He also writes the condition under which he would stop. That last piece is its own card; see related: Setting Kill Criteria Before You Start Something New.

References

• Kahneman, D. and Lovallo, D., "Timid Choices and Bold Forecasts: A Cognitive Perspective on Risk Taking," Management Science (1993), on the inside and outside views
• Flyvbjerg, B., work on reference class forecasting for large projects
• See related: Base Rates: The Number You Should Start From
• See related: Running a Premortem Before a Big Commitment
• See related: Outcome Bias: Judging a Decision by How It Turned Out