Guide
When a Site Is Losing Money: Fix, Shrink or Close
A site below water has three outcomes, not two. The commercial read that separates a demand problem from a conversion problem from a cost problem decides which.
A site losing money has three available outcomes: fix it, shrink it, or close it. Which one is correct depends on a question the P&L does not answer, which is whether the site has no demand, has demand it cannot convert, or converts its demand adequately on a cost base that its volume will never cover. Those three sites produce the same negative contribution line and call for three different decisions. The most expensive error in a multi-site platform is treating the second as the first, because a site with unconverted demand is the one case where closing destroys value that was actually there.
This guide is written for the sponsor and the CEO at the point one location is below water and the question has reached the board. It covers the decision and stops there. What follows a decision to close, meaning which customers transfer and what has to happen for them to, is covered in closing or merging a location. A site inside its opening ramp is not this case: a young location missing its plan is read against the curve it was underwritten on, covered in the new-location ramp curve, and treating a ramping site as a failing one closes locations that were on schedule.
The P&L gives you the symptom and none of the cause
Most platforms arrive at this decision with a site-level contribution report and a management view. The report is accurate and it is not a decision input. Negative contribution is consistent with all three causes, and nothing in the financial presentation separates them, because revenue and cost are outputs of the commercial process rather than descriptions of it. This is where the available outside material stops too: per-location profitability analysis and multi-site benchmarking identify which site is the problem, a different job from deciding what to do about it.
Three sites, one line in the report
A site with no demand produces inquiry volume, measured against the population and competitive set it serves, materially below what comparable sites produce. The market is thinner than underwritten, or a competitor already holds it. Everything downstream can be run well and the site still loses money, because there is not enough work in reach to cover the cost of being open.
A site with demand it cannot convert produces inquiry in line with its peers and loses it between the first call and the signed job. Quote volume is low against inquiry, quote-to-won is low against quote, or the jobs it wins are the small ones. The demand exists, somebody is serving it, and the site is not.
A site with a cost problem produces inquiry and converts it at the platform’s rate, and the economics still do not work. Revenue per job is structurally lower, or cost to serve structurally higher, usually through drive time, facility, wage market or a mix that skews to low-margin work. It is the only one of the three that is genuinely a finance question, and the least common in practice.
The read that separates them
The distinguishing evidence sits in systems the platform already has. Inquiry volume per site, normalized to the served market rather than compared raw, separates a demand problem from everything else. Comparing a dense suburban site to a rural one says nothing. Inquiry per thousand households in the service area, or per commercial account in the addressable set, says whether the phone is ringing at the rate the market should support.
Quote rate and quote-to-won, against the platform average, separate conversion from cost. A site taking inquiries and not producing quotes has a response, staffing or scheduling failure. A site quoting at volume and losing is priced wrong for its market or losing on something other than price, and the lost-reason field usually says which.
Revenue per job and job mix separate a cost problem from a conversion problem. A site winning at the platform rate on jobs that average materially smaller is serving a different market than the one underwritten, and the question in front of you is the size of the site rather than its existence.
Capacity utilization and lead response time, read together, catch the case that looks like weak demand and is not. A site that cannot answer its calls, or books far out when its peers book inside a week, produces the inquiry pattern of a weak market because the market stops calling. That failure mode is covered in selling into a capacity ceiling, and it is worth ruling out before concluding the demand is not there.
The read should produce one named cause. If it produces two, the sequence matters, because fixing conversion at a site with no demand buys nothing.
What a shrink actually means
Shrink is the option most platforms do not consider, and it is usually described wrongly when they do. Shrinking a site is not trimming its overhead. Cutting cost while holding the commercial footprint constant is a fix, and the version that works least often, because the cost base is mostly people and facility and neither comes out in slices.
A shrink is a commercial narrowing. The site keeps operating and stops trying to serve everything it currently serves. That is one of four moves: narrow the service line to the work the site actually wins and sell the rest from elsewhere, narrow the service area to the radius where drive time and win rate both hold, narrow the hours to the demand that exists, or stop quoting the segment the site loses anyway. The point is a site solvent at a smaller size rather than one chasing the size it was planned for. It fits two causes cleanly: a cost problem driven by mix or drive time, and a site with no demand in part of its area and adequate demand in the rest.
What a shrink costs gets understated. The platform’s promise narrows there, which affects national accounts and referral flow that assumed coverage, and route density falls in the area given up, raising cost to serve for what remains. A shrink is a decision to be smaller, not a softer version of doing nothing.
The order in which the three options foreclose each other
The three sit in an order, and taking one closes others.
A fix preserves everything. A site that is fixed can still be shrunk or closed later, and what is learned while fixing it improves both decisions. That is the argument for trying a fix first, and it holds only when the fix is tied to the named cause and to a date. A fix without a stated cause and a stated test is a delay, and a delay burns the same cash the close was meant to stop.
A shrink preserves the close and can foreclose the fix. Giving up a service line or area usually means releasing the crews, equipment or licenses that made it possible, and the local market’s belief that the site does that work goes with them. Re-entering an area the platform withdrew from is harder than entering it the first time, so a shrink is not a holding action while the real question is decided.
A close forecloses everything and its cost is one-way. The retainable share of the book is never all of it, the staff do not come back, and the market position does not wait. Against that, it is the only option that reliably stops the cash loss, and platforms that handle this badly are rarely the ones that close too fast.
Two dates drive the sequence more than the analysis does: the lease or facility commitment, and the point at which the cash loss becomes material to the covenant position. A decision that ignores them produces an option that was not actually available. A site gets one fix attempt against a named cause with a date attached. A second attempt at the same cause is a close that has been deferred, and it should be called one at the time.
What to require before the decision goes to the board
Four things, and a decision missing any of them is not ready. A named cause, stated as one of the three and supported by the numbers that distinguished it. A dated test, in commercial terms rather than a financial improvement that lags the cause by a quarter. The cost of being wrong in each direction, asymmetric and rarely written down: closing a convertible site costs the contribution it would have produced plus the market; holding a dead site costs cash on a known run rate. And a named owner, the platform’s commercial lead, not the general manager whose job is the subject of the decision.
The site that should stay below water on purpose
Applied mechanically, the framework above will close some sites that should stay open. A site that exists for route density, for coverage a national account requires, for a license the platform needs in order to bid, or as the entry point to a market the thesis depends on, is carrying a cost the platform chose. State what that site is for, set the number it is expected to hold rather than the one it was planned to earn, and report it as a strategic cost instead of a performance failure. What does not work is leaving it in the report as an underperformer, month after month, with no decision.
Frequently asked questions
How do you decide whether to fix, shrink or close an underperforming location?
By establishing the cause first. Negative contribution is consistent with no demand, with demand that is not converting, and with a cost base the site’s volume will never cover, and the three call for different decisions. Inquiry volume normalized to the served market, quote rate and quote-to-won against the platform average, and revenue per job against mix separate them. The decision follows the cause, not the size of the loss.
What does it mean to shrink a location rather than close it?
It means narrowing what the site commercially tries to do, not trimming its overhead. The usual moves are narrowing the service line to the work the site wins, narrowing the service area to where drive time and win rate hold, narrowing hours to the demand that exists, or dropping a segment the site loses anyway. The aim is a site solvent at a smaller size.
How long should you give an underperforming site before closing it?
Long enough for the named cause to respond, set by the site’s own cycles rather than a standard interval. A conversion fix shows up within a quoting and sales cycle. A demand fix carries a media and reputation lag. A cost or mix fix shows in the next full month of service. Set the test against the cause, attach a date, and treat a second attempt at the same cause as a deferred close.
Is a site losing money always a cost problem?
No, and it is the least common of the three. Most sites that look like cost problems are conversion problems wearing a cost problem’s numbers: low revenue per job because the site wins only small work, or high cost to serve because the schedule is loose rather than the wage market wrong. A genuine cost problem shows platform-level inquiry and conversion alongside structurally lower revenue per job, and the answer is usually a shrink.
Who should own the decision about a losing site?
The platform’s commercial lead should own the read and the recommendation, with the CEO owning the decision and the sponsor approving it where capital or the covenant position is affected. The site’s general manager should supply evidence and not own the recommendation: the decision is partly about that role, and leaving the analysis with the site puts the person closest to the data in the position of arguing for their own continuation.
Have a revenue problem the board is asking about? Start a conversation.