Two triggers, either one sufficient: replace when a unit's annual repair spend crosses roughly half its replacement cost, and replace when the unit's next failure would kill a booked Saturday. The first is accountant math, the second is operator math, and the second one matters more in a business where the gear's real job is making promises to families keepable. None of it works without one habit: a repair log per unit, because untracked repairs feel small forever.
The 50 percent rule, and why it needs a log
A cannon that cost 3,000 dollars and ate 1,600 in repairs this season is telling you something, but only if someone added it up. Repairs arrive one at a time, each individually reasonable, a pump here, a seal kit there, and the sum stays invisible without a log. The log is one line per incident: unit, date, what failed, what it cost in parts and hours, whether it happened at the shop or in the field. Five minutes a month, and suddenly the fleet has honest report cards.
When a unit's card shows repair spend approaching half of what a new one costs, the repairs have stopped being maintenance and started being installment payments on a machine you don't get to keep. That's the accountant trigger.
The Saturday trigger
The sharper question for this business: if this unit dies at 1pm on a booked Saturday, what happens? A failure that means a swapped backup and a ten-minute delay is a repair-side answer. A failure that means a canceled party, a refund, and a family's ruined birthday is a replace-side answer, almost regardless of the repair math, because the log can't price the review that says the foam never worked.
This is why the same component earns different answers on different fleets. The operator with three cannons can run one aging unit into the ground, since a field failure demotes to inconvenience. The operator whose single cannon is the entire business is holding a different asset: that unit failing costs a full Saturday of revenue plus reputation, and it should be retired earlier, or backed up sooner, than the accounting alone suggests.
The middle path: replace the parts that fail, keep the frame that doesn't
Most party equipment is a durable body wrapped around a handful of consumable organs: pumps, seals, hoses, switches, wheels on the game gear, blowers on the inflatables. The economical fleet strategy is usually aggressive on organs and patient on bodies: stock the known-failure parts, swap them proactively in the off-season rather than reactively in June, and let the frames run for years. The log tells you which parts are the organs, because the same failure showing up across units is a stocking list writing itself.
The off-season rebuild, every unit opened, worn organs swapped, everything tested, is the cheapest insurance in the business: an hour per unit in February prevents the July failure that costs a party.
When replacement is actually an upgrade
Sometimes the replace decision isn't about the old unit's health at all. Newer equipment that sets up faster, runs quieter, or foams harder changes the product and the schedule math: fifteen minutes off setup, across a season of multi-party Saturdays, is real capacity. Run that math when replacing anyway, because the answer sometimes justifies replacing early: beyond a healthier machine, you may be buying a shorter buffer and a fourth bookable slot.
And keep the retired unit if it still limps: a semi-retired cannon on the shop shelf is the backup that turns future field failures into ten-minute swaps, which quietly moves every other unit's Saturday trigger later. Old gear's last job is making the fleet braver.
The line to remember
Repair math is about the unit. Replace math is about the Saturday. Log the first, respect the second, and let old gear retire into being the backup that protects both.