Housing policy should rest on data.
Case closed. Problem solved?
Seattle has laws to protect renters, inspectors to enforce them, and a public database that tracks what happens when something goes wrong. Until you ask what “Closed” actually means.
STLCA examined 56,258 rental-housing records. 50,721 of them, more than 90%, are marked Closed or Completed. That status alone does not show that a repair was made, an unsafe condition was corrected, or a tenant was protected.
Closed/Completed does not mean Corrected/Protected.
The dates were map coordinates.
Until July 2025, the public field labeled LastInspDate did not show when a property was inspected. It showed latitude and longitude.
These are the records Council members, journalists, attorneys and renters use to see how housing laws are enforced. Inspection activity cannot be measured from a field that does not publish dates.
233,543 records. One question: what happened next?
We traced the complaint and enforcement records in the City’s open data from 2003 to 2025, published by SDCI, the department that enforces Seattle’s building and housing codes. Then we separated the 56,258 that concern rental housing.
Each section below shows one finding, in the City’s own numbers.
Complaints doubled. Formal findings fell.
When an inspector confirms a violation, SDCI issues a Notice of Violation: a formal finding with a deadline to fix the problem.
From 2017 to 2025, complaints rose from 7,768 to 15,973 a year. Notices fell from 1,673 to 969. Enforcement is also uneven by topic: land use and weeds draw most citations (10,732 and 7,270 of 20,340), and construction draws the most notices (8,470).
Both lines count separately opened records, not case-matched enforcement results.
2003 is a partial year: the data begins in November 2003.
A ratio of notice records to complaint records is not a matched complaint-to-enforcement rate.
Categories use the category label in the City record (RecordTypeDesc). Every record is counted once.
No category listed holds every tenant relocation (4,045) and unfit building (318) record, plus 19,736 complaints, 8,514 notices and 1,284 citations.
Landlord/Tenant (any label) counts every record whose label names Landlord/Tenant, alone or with another category: Landlord/Tenant alone (17,413), Emergency + Landlord/Tenant (2,776), Land use + Landlord/Tenant (83), Construction + Landlord/Tenant (26) and 8 smaller labels (27).
Emergency only is the single label Emergency. Blank labels are counted as No category listed.
Other and combined labels covers 58 other labels, led by Emergency + Vacant building (1,963), Tree (1,926), Shoreline (1,285) and Construction + Tree (544).
Landlord/Tenant (any label), 20,325 records, is not the 56,258-record landlord–tenant selection used in sections 02 to 07.
Closed/Completed is an administrative status. It does not show that a violation was corrected.
What the rental-housing records hold.
A noise complaint is not a broken heater. The 56,258 rental-housing records are not all complaints; they span five record types.
Most are renter complaints (40,540). The rest are Notices of Violation (9,961) and tenant relocation, citation and unfit-building records. For three in four of those complaints, 30,374 of 40,540 (74.9%), the public record shows no inspection date.
The selected 56,258 records are not 56,258 complaints. They are five record types.
The inspection-date rate is complaint-only. Other record types have different documentation workflows and are not part of it.
All-code records and landlord–tenant records are different populations.
In 2018, the inspection record went quiet.
Sort renter complaints by the year they were opened. Of those opened in 2017, 34.5% show no inspection date. Of those opened in 2018, 80.8% do, and every year since stays above 80%.
The blank is not limited to renters. Across all code types, 92,564 of 233,543 records have no date in that field.
Grouped by the year a complaint opened, in the later historical dataset. Not a snapshot of what the field displayed in 2018.
A blank last-inspection date is not proof no inspection occurred.
The 2018 opening-year split and the July 2025 date-column correction are separate questions.
Two problems, not one.
STLCA saw coordinates in the date field in March 2025. Dates for past records appeared in July 2025, and only then could dated and undated complaints be compared year by year. That is how the 2018 break came to light.
They are separate questions. The coordinates are about how the field was published. The 2018 break is about which complaints have a date at all.
The 2018 opening-year split and the July 2025 date-column correction are separate questions.
This is not just paperwork.
We sorted complaint text by topic: housing conditions, paperwork and tenant-protection issues, both, or other. Not every renter issue calls for the same response, so we checked complaints about physical conditions on their own.
Among closed complaints about living conditions, 848 of 7,035 opened before 2018 show no inspection date. For those opened from 2018 to 2025, it is 8,226 of 10,954.
Topic matches do not establish legally verified violations.
The habitability comparison uses a separate strict text-match subset; 6,925 of its 9,074 missing-date complaints have a review-only housing-relevance flag.
Every record type closes at about the same rate.
Complaints, Notices of Violation, tenant relocations, citations and unfit-building records are all marked Closed or Completed at 87.1% to 91.6%.
The published record gives no closure reason, so a repair, a finding of no violation and an administrative closure all look the same.
Closed/Completed is an administrative status. It does not by itself establish verified repair.
How often each closure reason occurs is not published in the record.
The public record does not link a complaint to its violation.
We matched them ourselves, by address and description. 6,240 of 9,961 notices (62.6%) trace back to a complaint, covering at least 5,323 complaints. The other 3,721 could not be matched this way.
Unmatched does not mean nothing was done. It means the public cannot follow the trail.
Method: address and description matches. Not exhaustive; no exact conversion rate.
Unmatched means not established by this method, not necessarily unrelated to rental housing.
A rental certificate is not a complaint response.
Under the Rental Registration and Inspection Ordinance (RRIO), Seattle registers rentals and inspects them on a schedule. A building can pass that inspection and still draw renter complaints later.
Registration history and complaint outcomes sit in separate public records, so seeing both for one building means searching each.
RRIO certification is not a response to a subsequent complaint.
The schematic shows how the two records are structured. It does not count properties.
One building, records in several City systems.
At Addison on Fourth, 308 4th Ave S, the housing record told only part of the story. Fire Department dispatch logs, rescue calls and court filings filled in more.
The case study puts every record in date order and marks what each agency documented and what is still unverified.
The case study places records in date order. It does not draw conclusions from one number.
Cross-agency totals are not published until the record-level register is complete.
Counting complaints is not enough.
Every case should carry its full chain in one place: complaint, first response, inspection, finding, notice, deadline, repair, verification and the reason it closed.
STLCA puts the public record together for renters, reporters, attorneys, researchers and policymakers. We are asking City Council for complete, linked enforcement data.
Housing policy should rest on evidence the public can follow.
Pick a District
Each of Seattle’s seven Council districts has two towers on one scale: renter complaints and Notices of Violation. Hover or tap a tower to open that district’s dashboard or dossier.
Rental-housing records opened 2003–2025: 39,995 complaints and 9,892 Notices of Violation fall inside a district; 545 complaints and 69 notices have no district. Together: 40,540 complaints and 9,961 notices.
Publication boundary
All-code records and the landlord–tenant selection are different populations; the counts shown are historical, not a live SDCI feed. The selected 56,258 rows contain five record types. Inspection-date completeness on this page is reported separately for the 40,540 complaint records because the other record types have different workflows. A blank last-inspection date does not prove an inspection never occurred. Closed/Completed does not establish verified repair. Notice-to-complaint ratios are not case-matched conversion rates. Annual graphs use opening year, including years before the July 2025 field backfill.
Sources used on this page
- Seattle open data portal: Code Complaints and Violations (Built Environment)City of Seattle, SDCI. Used for: All-code records, 2003–2025 (sections 01, 03).
- landlord-tenant-records.csvSTLCA landlord–tenant selection, 56,258 rows, opened 2003–2025. Used for: Record types, statuses, inspection-date field, district towers (sections 02, 06, Pick a District).
- STLCA_STORY_DATA_AND_EVIDENCE_INDEX_2026-11-15.jsonSTLCA story data and evidence index. Used for: Annual series, category counts, living-condition and matching figures (sections 01, 03, 05, 07).
- report-notices-matched-to-complaints.htmlSTLCA Notices Matched to Complaints report. Used for: Notice-to-complaint matching (section 07).
- citywide-map.htmlSTLCA citywide complaint maps. Used for: Geographic views (section 05).
landlord-tenant-records.csv · 56,258 rows · SHA-256 b55cd317002c16c0fc4b5271c324af8b6fb01b11af79d8b93cc41a818db49076