I stayed at a Sonder in Mexico City in 2025. The apartment was clean, the app worked, and the stay was better than I expected. That experience made the company’s failure more interesting to me. A good guest product could not cover the obligations behind it.
On November 10, 2025, Sonder announced an immediate wind-down after Marriott ended its licensing agreement. Sonder did not file for bankruptcy in October. Its SEC filing says the Chapter 7 petitions were filed November 14. Chapter 7 means liquidation, a sharper outcome than a restructuring plan.
Sonder’s model needed more than a booking app. It took on leases, furnished rooms, handled cleaning, and supported guests. Rent and operating staff still cost money when occupancy falls. The company controlled the guest experience more closely than a marketplace, but it also carried more of the cost and risk. That trade was visible even in a stay I liked: the clean room and reliable service required local people and a building somebody had to pay for.
Marriott’s agreement was supposed to bring Sonder distribution through a much larger hotel network. Marriott terminated it in November, saying Sonder was in default. I would not claim Marriott walked away because it knew the unit economics were impossible; the public statement does not establish that. The termination did remove an important route to bookings at a moment Sonder had little room left.
The lesson I take from my stay is uncomfortable: customer satisfaction is necessary, but it does not pay a fixed lease on an empty night. The useful numbers for any managed-apartment business are occupancy, revenue after distribution fees, labor per occupied unit, and lease commitments through a downturn. Sonder built a product I would have booked again. Its balance sheet could not wait for that next booking.