March 1, 2026
By the HalfKey team
30 vs 90 nights: the pricing math behind the discount
Most Tokyo monthly mansion operators quote a cheaper per-night rate at 90 nights than at 30. The discount is real on the tenant's side. On the operator's side, most of it is not a marginal-cost saving. It is the price of locking down a slot that would otherwise risk vacancy.
The 90-night rate on a Tokyo monthly mansion looks 8–15% cheaper per month than the 30-night rate. On a ¥210,000 baseline that's a ¥30,000 to ¥80,000 saving across the full stay. Sam's piece on what four friends actually did covers the reader-side risk math. This one covers the operator's spreadsheet. What the discount actually represents, when it tracks marginal cost, and when it is a vacancy-management mechanism dressed as a customer benefit.
The short version: the discount has two components, only one of which is a real cost saving. The other is the operator buying down vacancy risk by handing the tenant a number that looks generous. Knowing which is which tells you whether the 90-night booking is a genuine win or whether you're underwriting somebody else's calendar problem.
Start with the 30-night tier. A residential 1K of around 22m² in Suginami, Setagaya, or Nakano typically lists at ¥165,000 to ¥230,000 for 30 nights. That figure is all-in before utilities and cleaning. It works out to ¥5,500 to ¥7,700 per night. The operator's gross on that month, after paying the Building Co. ¥125,000 of sublease (see master-lease structure separately), one cleaner at ¥18,000, and platform fees of around ¥10,000, lands near ¥85,000. That ¥85,000 is the 30-night tier's contribution margin.
Per-night yield is high. Per-stay risk is also high. The tenant can leave at day 30 with no penalty, the slot reopens, and the operator is back to advertising and waiting. On a 12-month calendar of consecutive 30-night stays, the operator runs 12 cleanings, 12 onboardings, and 12 listing-slot exposures. Each gap between stays is unrecouped sublease. A 5-day gap on a residential unit is roughly ¥21,000 of sublease the operator paid for nights they did not sell.
Now the 90-night tier. The same unit at ¥185,000 per month for a 90-night booking generates ¥555,000 of rent across the stay. Operator costs over those 90 days break down as follows. Sublease: ¥125,000 × 3, or ¥375,000. Cleaning: ¥18,000, one stay not three. Onboarding: maybe ¥8,000 of staff time, again one stay not three. Platform fees: about ¥15,000 for one booking. Operator gross: ¥139,000 across 90 nights, or ¥1,544 per night.
Compare that to the 30-night tier yielding ¥85,000 per month, or ¥2,833 per night across consecutive bookings if the calendar stays full. The per-night gross at the 30-night tier is 1.8× the per-night gross at the 90-night tier. The operator is not richer at 90 nights. They are surer.
The discount the tenant sees is ¥75,000 across 90 nights, versus three stacked 30-night bookings at ¥210,000 each. That is the operator giving up about ¥75,000 of nominal revenue. They get back ¥36,000 of marginal-cost savings: two cleanings avoided at ¥18,000 each. The remaining ¥39,000 of "discount" is the operator paying the tenant to remove vacancy risk. It is not a cost saving. It is a hedge premium.
This is where the operator-incentive picture gets clean. The 30-night tier is a high-yield, high-variance product. The operator wins big when the calendar stays full and absorbs the loss when it does not. The 90-night tier is a low-yield, low-variance product. The operator gives up upside in exchange for guaranteed occupancy of the slot.
The portfolio math forces the choice. An operator running 80 units in Suginami and Nakano carries roughly ¥10 million per month of sublease obligation. Their booking team is not optimizing for the highest per-night rate. They are optimizing for the fewest empty nights. A 90-night booking removes 90 units of empty-night risk in one signature. The discount is the price they pay for that certainty. It comes out of expected revenue, not out of cost.
The portion that does represent real marginal cost is straightforward. Cleaning at ¥18,000 per turn is the largest line. One stay instead of three saves two cleanings, ¥36,000. Onboarding labor is real but smaller, maybe ¥5,000 to ¥10,000 in staff time per check-in. Platform fees are stay-based, not night-based, so a single 90-night booking on Suumo or GaijinPot Apartments incurs one platform fee instead of three. Add it up: roughly ¥40,000 to ¥55,000 of genuine marginal-cost savings on a 90-night stay versus three 30-night stays. Anything beyond that is hedging.
The cancellation grid is the other side of the same trade. A 30-night booking has near-zero exit cost: you complete the month, you leave, the operator's exposure ends with the cleaning fee they already collected. A 90-night booking past day 14 has a much steeper price. Typical mid-term operator grids charge three things at once. An early-termination fee of roughly one month's rent (¥185,000). The current month forfeited. Pre-paid future months refunded only against 30 days written notice. On a ¥555,000 90-night stay, walking at day 40 can cost the tenant ¥185,000 to ¥370,000 against an original ¥75,000 of saving.
That is the operator's optionality cost being charged back to the tenant who declined to pay it upfront. The 90-night discount is the operator selling certainty cheaply. The cancellation grid is the operator buying it back at retail when the tenant tries to undo the deal. Both halves of the trade are priced. Sam's piece walks through what this looks like from the lived side.
The 2x2 falls out cleanly. Two axes. Confidence-in-fit: do you already know this ward, this unit type, this commute works for you. Budget sensitivity: does ¥75,000 across 90 nights move the dial.
High-confidence-in-fit, budget-sensitive: book 90 nights. The discount is genuinely yours because you are not paying the optionality value back at day 40. The hedge premium the operator is paying to lock you in becomes a real win.
Low-confidence-in-fit, budget-flexible: book 30 nights and renew. The optionality has measurable value to you, and the discount the operator quotes does not capture it. You are paying ¥833 per night extra (the ¥75,000 spread over 90 nights) for the right to walk after month one. That is cheap insurance against the cancellation-grid scenario.
Low-confidence-in-fit, budget-sensitive: still 30 nights. The asymmetric tail of "wrong unit at 90 nights" wipes the discount roughly six times over.
High-confidence-in-fit, budget-flexible: indifferent on saving. A 90-night booking does favor the operator's calendar, though, and that sometimes earns goodwill on cleaning-fee waivers or unit upgrades. Not a primary reason. A real one.
The breakeven point is a confidence threshold, not a yen number. If you have already lived in your target ward for 14+ days on a prior visit, the 90-night rate beats stacking renewals. The optionality you are giving up has low expected value. The ¥75,000 saving is mostly real. If you have not lived there yet, the optionality has measurable expected value. Your future self will want the right to switch wards roughly 30% of the time, going off four observed first-stays. The discount math under-prices that.
Sam's piece on what those four friends actually did covers the reader-side decision; this one covers the operator's spreadsheet. The two halves are the same trade priced from opposite ends. The operator's discount and the tenant's optionality are the same number, paid in different currencies.
The next time you see "save ¥75,000 by booking 90 nights" on a Tokyo monthly listing, split the number in two. Half of it is two cleanings the operator did not have to do. The other half is the operator paying you to take their vacancy risk off the table. Then check whether you actually want to take it.