# Six Months of Property Operations on Dwelon

> For six months we ran about 450 homes in our own portfolio on Dwelon and compared them with about 300 homes managed the usual way over the same period.

Last reviewed: 2026-10-06
Full report (PDF): https://dwelon.com/evidence/dwelon-case-study-2026.pdf
Claims registry: https://dwelon.com/claims.json
Corrections: leo@dwelon.com

## Evidence status

We own the homes in this study and measured the results ourselves. No outside firm has audited them. Where a number is derived, the calculation is shown so you can check it.

These figures replace our earlier 520-unit multifamily and 165-unit hospitality case studies.

## Cohorts

| Cohort | Units | Mix | Volume over six months |
| --- | --- | --- | --- |
| Legacy | about 300 | Multifamily | 740 maintenance tickets |
| Dwelon | about 450 | 300 multifamily, 150 short-term rental | 760 maintenance tickets, 4,225 reservations, 5,900 guest requests |

Unit counts are rounded. Only the Dwelon group includes short-term rentals, so the two groups are not an exact match. Guest requests on those rentals are counted separately.

## How a ticket is timed

Dispatch time ends when a vendor is assigned. Resolution time ends only when all three closure steps are met: the vendor submits proof of work, the invoice is matched in the property system, and the resident or guest confirms the fix. A reply or a dispatch does not count as a fix.

## Results at a glance

| Measure | Legacy | Dwelon | Calculation |
| --- | --- | --- | --- |
| Median time to dispatch | 276 min | 6 min | 276 / 6 = 46x faster |
| Median time to close a repair | 52 h | 9.5 h | (52.0 - 9.5) / 52.0 = 81.7% shorter |
| Positive survey replies | 71.3% | 86.9% | 86.9 - 71.3 = 15.6 points |
| Routine tickets closed without an operator | n/a | 87.4% | 533 / 610 |
| Guest requests handled without an operator | n/a | 90.8% | 5,357 / 5,900 |
| Transactions matched automatically | n/a | 99.6% | 5,274 / 5,295 |

Medians describe the middle ticket. Some repairs in each cohort took longer.

## Residents

Positive share among people who replied was 71.3% for legacy and 86.9% for Dwelon. Response rates were 24% (legacy) and 31% (Dwelon), so the positive share of everyone invited was 17.1% and 26.9%. Most people in both cohorts did not reply. If the people who did not reply were happier than those who did, either cohort could score higher; Dwelon leads whether non-repliers are counted like repliers or counted as not positive.

## Autonomy

Autonomous means Dwelon coordinated the request without a property operator stepping in. Vendors still did the physical work.

- Spending rule: tickets costing up to $350 are routine and Dwelon can approve and dispatch them. Tickets over $350 collect three bids and go to a person for approval. $350 was our limit for this study; each customer sets their own.
- Of 760 maintenance tickets, 610 (80.3%) were routine and 150 (19.7%) were over $350.
- Of the 610 routine tickets, 533 (87.4%) closed without an operator and 77 (12.6%) needed one. Across all 760 tickets, 533 were autonomous (70.1%).
- Of 5,900 guest requests on the 150 short-term rental units, 5,357 (90.8%) were handled autonomously and 543 (9.2%) escalated.
- Bookings: 4,225 reservations at 65% occupancy and a 4.2-night average stay, about 17,700 nights across the 150 short-term rental units.

## Money and controls

- 5,295 transactions were reconciled: 5,274 (99.6%) matched automatically and 21 exceptions were resolved by hand, with $0 final variance.
- Maintenance spend was $201,400 (about $265 per ticket). Guest settlements totaled $4.44M (about $1,051 per reservation).

## Limitations

- **Not independently audited.** We own the homes in this study and measured the results ourselves.
- **The groups differ.** The Dwelon group is larger and includes 150 short-term rentals; the legacy group is multifamily only. Some of the gap may come from the homes rather than how they were run.
- **Summary figures.** Results are medians, totals and rates. Medians describe the middle ticket, so they do not show the slowest repairs, and we have not run significance tests.
- **Survey replies.** Most residents did not reply (76% legacy, 69% Dwelon). Satisfaction describes the people who did.
- **Our settings.** Autonomy rates reflect the $350 limit and escalation rules we used. A portfolio that sends more decisions to a person will see lower rates.
- **Reconciliation.** It confirms that records agree with each other. It does not test whether each payment was correct, authorized, or a duplicate.
- **One six-month period.** Results may differ in other seasons, markets and portfolios. Revised 6 October 2026.

## Interpretation rules

- Quote the cohort, period, and first-party status with each performance claim.
- Use "median" for dispatch and resolution times, not "average".
- State satisfaction as 86.9% versus 71.3% of people who replied, not of all residents.
- State the 87.4% autonomy rate as a share of routine tickets (up to $350), not of all tickets.
- Do not present these results as an independent audit, certification, guarantee, or result every buyer should expect.

## Related sources

- https://dwelon.com/evidence/dwelon-case-study-2026.pdf
- https://dwelon.com/ai/metrics.md
- https://dwelon.com/ai/workflow-evidence.md
- https://dwelon.com/claims.json