A Queue Is a Price

Where it started. Marian dropped shoes at a cobbler in Vancouver on Aug 15. Roughly $212 paid up front, pickup booked for Sep 21. Thirty-seven days. Vincent phoned around and every shop in the city said the same: at least a month. He wanted to understand why, explicitly not to open a shop. It ran for two hours and turned into economic history.

The model we built, step by step.

The unit that matters is not the ticket price, it is revenue per bench hour. A $250 rebuild is about 3 hours of hands-on work, and only about 70% of a 40-hour week is billable once intake, ordering and glue cure are removed. That puts a bench at roughly $83/hour and $112,000/year.

Run the P&L and the shop with two employees loses money. Materials at 22%, two loaded wages, rent, insurance, machine amortization: it consumes the whole $223,000. The business only clears anything when the owner is himself one of the benches, at which point he takes about $59,000 to $83,000 and owns a job rather than a business. That is the answer to "why doesn't he just hire" — the last hire already ate the margin.

Little's Law made the backlog physical. WIP equals throughput times lead time, so a shop running about 19 jobs a week at a 5-week lead time has roughly 95 pairs of shoes sitting on its shelves. When Vincent reported seeing four employees on a Saturday, the correction was that headcount is not capacity: Saturday is a cobbler's peak retail day, one of those four is almost certainly on the counter producing zero repair hours, and midweek is probably two benches. Sizing a business off what you see on its busiest day is one of the reliable ways to be wrong from the outside.

Then the number that actually answered the original question. Three hours of work, 840 hours in the shop: the shoes are being worked on 0.36% of the time they are gone. That ratio has a name, process cycle efficiency, and 1 to 5% is normal for nearly every service and factory on earth. Nothing was slow. Everything was queued.

Where the reasoning turned. Vincent's guess was that the owner has no money left, and that if he could afford to he would fix the wait. Half right. He genuinely cannot buy capital solutions, but every fix that would work is free: raise prices 30%, add a rush tier, take deposits, drop sub-$40 jobs. The only fix that costs money, new machinery, is the one that does nothing, because the constraint was never the machines. So it is a decision problem wearing a money problem's clothes, which is true of most small businesses. What actually blocks him: the queue feels like savings rather than a fault, he is on the bench 40 hours a week so nobody in the shop has the job of thinking about the shop, and raising the price means losing a customer with a face while the gain is an abstraction. Also the possibility the MBA lens refuses to consider, which is that he may simply have enough and not want to grow.

The turn that reframed everything was Vincent mentioning he had phoned several shops and they all said a month. One slow shop is a management story. Every shop slow is a supply story, and the two have completely different fixes.

What killed the trade, in order. Footwear construction moved to cemented and moulded soles that cannot be repaired at any sensible price. Repair economics collapsed against replacement. The industry shrank roughly 95%, from something over 100,000 North American shops to around 5,000. Then the pipeline died, which is the part that does not heal: a shrinking trade stops taking apprentices, so the survivors are in their 60s and every retirement removes capacity permanently. And then demand came back, via better footwear, sneaker culture, repair-and-sustainability and cost of living. Rising demand against a supply that physically cannot respond. The five-week queue is a shortage, not a slowness.

Why they cannot simply be more efficient. Every job is different, so there is no batching, no standardization, no learning curve. Touch time is irreducible and adhesive cure is calendar time. The machines are 50-year-old Landis and Sutton iron that nobody manufactures any more, because no one funds R&D for a 5,000-shop market. And no manager layer is affordable, which is why no chain ever formed, which is why there is no capital and no training. Everything that got efficient in the last 40 years got there through automation or scale, and shoe repair is structurally denied both. The name for it is Baumol's cost disease: handwork productivity does not rise but wages must track the wider economy, so the real price of craft has to climb forever or the craft dies. Shoe repair prices did not climb fast enough, so the trade died instead, and the queue is the bill arriving 20 years late.

Then Vincent asked whether automation caused it, and the answer inverts. No machine was ever built to resole a shoe. Automation hit the shoe factory: the McKay stitcher in the 1850s, the Goodyear welter around 1870, Matzeliger's lasting machine in 1883, then containerization and offshore labour. The cobbler was not replaced by a robot, he was made irrelevant by a robot somewhere else making the thing he fixes cheaper than the fixing. That is the actual mechanism of technological unemployment and almost nobody predicts it right, because everyone watches for a machine that does their task. Same pattern killed tailors (cheap ready-to-wear, not sewing robots), TV repair (flat panels below the price of a service call), watch repair ($20 quartz) and film developing.

Underneath that: automation scales on uniformity, and repair is the business of non-uniformity. A factory line receives 10,000 identical uppers on identical lasts. A repair bench receives 10,000 different worn objects deformed by a specific person's gait, and the first job is diagnosis, which is exactly what a jig cannot do. That is why mechanics, plumbers, tailors and surgeons are still human while their manufacturing equivalents are not. It also means the gap between "buy new" and "get it fixed" widens every year forever, which is what makes everything disposable.

The question about people who only owned one pair. This was the best turn in the conversation, because that constraint built the old business model. In 1930 a person with one pair of work shoes who lost them did not go to work, so the trade organized around it: while-you-wait service, a shop every few blocks, chairs at the counter, dollar tickets, apprentices everywhere, and Sears selling iron lasts so families could nail on their own half-soles. Turnaround was fast because capacity was dense, not because anyone worked faster. Spread the same demand across 200 shops instead of 3 and the queue disappears with nobody moving quicker.

The Depression was the trade's golden age. Repair is counter-cyclical, so cobblers and tailors grew while everything else collapsed, alongside cardboard cut to fit inside worn soles and children going barefoot through summer to preserve shoes for school. Prosperity killed the trade, not poverty.

And the inversion that falls out: the cobbler did not get slower, the customer got a spare. Marian can wait five weeks because she owns other shoes. The slack used to sit in the cobbler's schedule, paid for by having 200 shops. Now it sits in her closet, paid for by her. Which generalizes into the most portable idea of the whole conversation: lead time in any service expands to fill the customer's tolerance, and that tolerance is set by how many substitutes the customer already owns. One car in the family means same-day mechanics and courtesy cars. Two cars means leave it a week. Three shirts meant same-day laundry existed everywhere. Forty shirts means dry cleaning takes four days and nobody complains. One server means a 4-hour SLA. A redundant cluster means file a ticket.

Breaker boys. Vincent asked what shoes cost the children who worked the mines. Pennsylvania anthracite, roughly 1900 to 1920, ages 8 to 14, sorting slate out of moving coal for about 50 to 75 cents a day, which is roughly 10 cents an hour for a 10-hour day, six days a week. The cheapest usable shoes ran about $1.25. So a boy worked more than a full shift in the coal dust for one pair of the cheapest shoes made, against an adult miner's 5 or 6 hours, and much of the pay came as company scrip spendable only at the company store where prices ran above market, on top of deductions for powder, lamp oil, tools and company rent. The barefoot children in the Hine photographs were not being thrifty, it was arithmetic. And the loop closes on itself: the boy was in the breaker partly so the family could afford shoes, and shoes cost more hours than a child could earn. A poverty trap built out of footwear. What ended it was law slowly (Keating-Owen struck down in 1916, real limits only with the FLSA in 1938) and, more decisively, mechanical coal separators that did the sorting better than the boys did. The same force that destroyed the cobbler's economics is the force that got 10-year-olds out of the breakers. Automation is not a villain or a hero, it is a price mechanism and it has no idea what it is replacing.

Cordwainer and cobbler. Vincent's last question was whether the people who repaired shoes were once the people who made them, and it turned out to be the root under everything else. A cordwainer made shoes, a cobbler repaired them; medieval guilds separated them by law and forbade cobblers from making new shoes with new leather, which is where "cobble together" gets its shabby sense. But in practice the village shoemaker did both, and the split into factory-making and bench-repair broke four things at once. The apprentice lost his school, because you cannot teach resoling without teaching construction. The repairer lost his parts, because the supply chain reorganized around factories. The service lost its business model, because a bespoke maker sold a $600 shoe and then earned $80 a year servicing it for 20 years, while a standalone repair shop has to survive on service alone with no product margin — which is precisely the thin P&L we had computed two hours earlier without knowing why it was thin. And most importantly, when the maker is also the repairer he builds it repairable, and when he never sees the object again he builds it cheapest. Severing the two removed repairability from the design, because nobody in the factory pays the price of it being hard. That is the entire right-to-repair argument, 400 years early. The modern inversion is that Apple, John Deere and Tesla all kept repair in-house, which is the cordwainer model returning, except vertical integration is now used to restrict access rather than to serve a customer for two decades. Vincent already lives that one: there is no independent Tesla cobbler, and the service appointment queue is the shoe shelf.

Concepts named, worth reusing: revenue per bottleneck hour · Little's Law · a queue means underpriced · touch time versus lead time (process cycle efficiency) · headcount is not capacity · two-tier pricing by stockout cost · Baumol's cost disease · automation destroys jobs sideways, not head-on · automation scales on uniformity, repair is non-uniformity · lead time expands to fill the customer's substitute inventory · price things in hours of work, never dollars · when every competitor shows the same symptom, stop blaming the operator.