Two work boots sit side by side in a store. One costs $60, one costs $180, and every frugal instinct in your body points at the $60 pair. Now fast-forward three years: the $180 boots have been worn about 300 days a year and are on their second resole, while the $60 pair gave out twice and was replaced four times total. Over those three years, the cheap boots cost $360 and the good boots cost $180. Per day on your feet, the expensive pair was half the price. Cost per use is the framework that makes that reversal visible in the store, before the money is spent — and it’s the single most useful buying tool I know, right up until it breaks.
The One-Division Framework
Cost per use (CPU) is exactly what it says: the price of the thing divided by the number of times you use it. A $50 coat worn 100 times costs $0.50 per wear. A $200 coat worn 100 times costs $2.00 per wear — unless you wear it 400 times, at which point it’s $0.50 too. Every purchase is a guess about the denominator, which is why the framework is as much about honesty as arithmetic: the expensive thing only wins if you actually use it, and the cheap thing only loses if it dies young.
Three worked examples, all arithmetic you can check in your head:
| Purchase | Cheap option | Quality option | Winner |
|---|---|---|---|
| Work boots, ~300 wears/year | $60 every 6 months → $360 / 3 yrs = $0.40/wear | $180 for 3 years = $180 / 900 wears = $0.20/wear | Quality, 2x better |
| Gym membership | $50/month ÷ 8 visits = $6.25/visit, but ÷ 3 visits = $16.67/visit, and ÷ 12 visits = $4.17/visit | Depends entirely on usage | |
| Chef’s knife | $40, replaced every 2 years = $20/year | $150, kept a decade = $15/year | Quality, barely — and it cuts better |
Notice what the framework did and didn’t do. It confirmed the boots case decisively, because usage was high and predictable. It refused to crown a knife winner (a $5-a-year difference is noise), which is itself useful — CPU’s job is to kill bad comparisons, not to manufacture drama. And on the gym it said the only true thing: the answer lives in your calendar, not the price list.
Where the Framework Earns Its Keep
CPU is most powerful in the purchases where sticker shock pushes you wrong. The pattern to memorize: the things you use daily deserve quality; the things you use rarely can be cheap. A mattress you sleep on 3,650 times over ten years, an office chair you sit in 2,000 hours a year, shoes that touch pavement 300 days a year, the kitchen tools that cook 600 meals a year — those categories have enormous use counts, which means quality is cheap per use even when it’s expensive per receipt. Meanwhile the pasta maker you’ve used four times since the wedding, the kayak that’s been on the wall since 2019, and the bread machine from the phase — those have terrible CPUs regardless of price, and the fix isn’t a better model, it’s an exit.
This is also the framework that saves you from the opposite failure: cheap things that generate hidden follow-on costs. A $30 suitcase that dies on trip three and mangles a $400 pair of shoes along the way has a worse CPU than the $140 bag that lasted a decade, but the sticker never told you that. When you catch yourself replacing the same $15 umbrella for the fourth time, you’re paying $60 for one umbrella with extra steps. There’s a famous name for this — the Vimes boots theory from Terry Pratchett’s Discworld, where the rich man’s $50 boots outlast ten rounds of the poor man’s $10 boots — and it’s worth reading in the original, because it’s really a post about how being cheap can be expensive. I’ve written the full version of that logic in buying quality vs buying cheap; this post is the tool, that one is the category-by-category application.
Where the Rule Breaks: The Honest Section
The “$1 per use is a good deal” rule of thumb that floats around personal finance is a fine starting heuristic, and it’s wrong in enough places that you need the exceptions taped next to it:
- Experiences don’t scale with repetition. A wedding, a family trip, a graduation — used once, “wasted” by CPU math, worth every cent. CPU assumes value comes from repeated use, and for memory purchases the model is just wrong. Don’t CPU your honeymoon.
- Safety gear has no CPU. A bike helmet, a car seat, a ladder — you’re not buying uses, you’re buying the tail risk. Buy these good and new; $0.30 per use is the wrong lens entirely.
- High-CPU purchases you keep anyway can be correct. The stroller used 40 times that kept a marriage functioning on airport trips: the counter is the number of uses, but the value isn’t. Some things buy convenience and dignity, not repetitions.
- The sunk-cost trap runs the rule backward. The nastiest failure mode: you own the $300 bread machine, so you keep “using” food you don’t enjoy to push the per-use number down. CPU is a buying tool, not a guilt engine. Once money is spent, the only question is what’s best going forward.
- Usage optimism is the original sin. Most CPU failures happen in the denominator, not the numerator. The rowing machine math assumed four workouts a week. Be brutal: use your actual last-90-days behavior, not your intentions, which is the same discipline that powers the subscription audit — same division, recurring charges instead of purchases.
The 15-Minute Version: Run It on One Decision
You don’t need a spreadsheet for this framework; you need it on the one purchase currently sitting in your cart. Five questions, written on a sticky note:
- How many uses did the last version actually get? (Past behavior, not hopes.)
- What does that make the price-per-use of each option on the table?
- What’s the cost of the cheap one failing early — replacement, damage, a wasted trip?
- Is this a repetitions purchase, a safety purchase, or a memory purchase? (Only the first one belongs in the CPU frame.)
- Could a used version buy quality at a cheap price? Tools, bikes, furniture, and gear are the classic categories here — the depreciation happened on somebody else’s receipt. That whole strategy is its own post: how used and thrift items accelerate your financial freedom path.
One more extension worth a sentence: the same division works on time, not just money. A gadget that saves 20 minutes a day is 120 hours a year; the real question is what those hours become. I run that version in the dollar-per-hour lens on high-value activities — CPU for purchases, hours-per-week for your calendar, same idea.
The Bottom Line
Cost per use won’t make every purchase for you, and it shouldn’t — its highest value is as a lie detector. It exposes the cheap things that are secretly expensive, the expensive things that are secretly cheap, and, most usefully, the purchases where you were about to pay for a fantasy version of yourself who uses things more than the real you ever has. Run the division before the purchase; run it again a year later on the things you own, just to calibrate your own denominator for next time.
Forward step: pick the one purchase you’re currently debating — in your cart, in a tab, in your head — and answer the five sticky-note questions in writing. If the per-use number embarrasses you, close the tab. If it surprises you in the good direction, buy it without guilt, because that’s the sound of a framework doing its job: not more spending, not less spending — accurate spending.
