For public housing authorities and voucher programmes
Waiting list position is the most consequential ordering in local government — years of a household’s life turn on it — and it is set by your board-adopted admissions plan and preferences, not by any system. Eligibility, continued occupancy, termination and inspection outcomes stay with your staff. What can be carried is the document chase, the scheduling, and reaching households whose phone number changed two years ago.
The waiting list is long enough that a household’s circumstances change several times before their name comes up. They move, their household composition changes, their income changes, their phone number changes twice — and the authority’s record of them is from the day they applied.
So when the name does come up, the letter goes to an address from three years ago. The household does not respond, is removed for non-response, and reapplies at the bottom. Nobody did anything wrong and a family lost three years.
Recertification is annual and document-heavy for households least able to assemble documents. Income verification from an employer who does not respond, a benefit letter from an agency on a different cycle, a birth certificate for a child added to the household.
Inspections are a scheduling problem with three parties — the household, the landlord and the inspector — and a failed inspection means a re-inspection, which means another scheduling round, during which a voucher payment may be suspended and a tenancy may be at risk.
And the reporting is federal, detailed and tied to funding, produced by an authority that is also running property management, a voucher programme, and a waiting list of several thousand people.
The authority’s contact record is from the application date and the wait is measured in years, so the notification that matters most in a household’s life is sent to an address that is systematically likely to be wrong.
That is the most consequential administrative failure in this domain and it is entirely a reach problem. It is not a decision anybody made and it is not visible in any report — the removal is recorded as non-response, which reads as a household that declined.
The first act is to measure it: for a prior list pull, how many removals for non-response followed a notification where every channel on file had already failed. Authorities are frequently startled, and it is the number that changes what the board thinks the list is.
What can then be carried is reach and completeness. Keeping contact details current between the application and the pull, through channels households actually use. Naming every recertification document at once rather than one at a time. Chasing the third-party verification. Scheduling an inspection across three calendars and confirming before the date. Assembling the federal report from data the authority already holds.
What never moves, at any tier: waiting list position or ordering; preference determination; eligibility, continued occupancy or termination determination; rent calculation; inspection outcome; and any scoring, ranking or prioritisation of households. Preferences are adopted by your board in an admissions plan, and ordering is the plan applied — not a system’s judgement.
Removals for non-response where every channel had failed — measured by count of removals preceded by a failed notification, which most authorities have never separated from genuine declines.
Households with current contact details at the pull — measured by share reachable on first attempt when their name comes up, against the prior pull.
Recertification documents requested one at a time — measured by count of separate requests per household, before and after.
Third-party verifications outstanding — measured by count and age of outstanding employer and agency verifications.
Inspection scheduling rounds — measured by count of scheduling attempts before three calendars align, and confirmations before the date.
Landlords leaving over administrative friction — measured by count of landlords exiting, with reason recorded, against the prior year.
Time to assemble a federal report — measured by elapsed hours from a report becoming due to submission.
any waiting list position or ordering; any preference determination; any eligibility, continued occupancy, rent calculation or termination determination; any inspection outcome; and any scoring, ranking or prioritisation of households by any characteristic or prediction. Preferences are adopted by your board in an admissions plan and the ordering is that plan applied — never a system’s judgement.
It reads and writes the housing management system for the administrative purposes named. It has no write path to waiting list position or preference fields, and no read access to the ordering logic is requested — proximity is the risk here more than anywhere else in this wave, so the access itself is absent.
Recertification requirement lists come from your own administrative plan and the federal programme rules as your staff already apply them. Nothing interprets a rule.
Nothing creates a second record of a household. In a programme where continued occupancy is at stake, a second record is a due process problem.
And household-facing surfaces are built for the population served — an old phone, a screen reader, large text, the languages your service area speaks, and a phone number that reaches a person.
Waiting list position determines years of a household’s life. It is set by preferences your board adopted in a public process, and applying those preferences is an act of your authority under federal rule and under fair housing law.
So the exclusion covers position, preference, ordering and any scoring that could inform one — and the access is absent rather than restricted, because a credential that could reach the ordering is a fair housing question regardless of whether it was used.
Household information here discloses income, household composition, immigration-adjacent facts and sometimes disability and domestic violence circumstances. It stays inside your tenancy, on your retention schedule, exportable by you, and is never training material at any tier. Confidentiality protections for households fleeing violence are handled under their own rules and default to the more protective reading.
And on assurance: an independent SOC 2 Type II attestation is in progress and no report exists yet.
The director should test the ordering exclusion by asking for the access list rather than the policy: confirm no credential reaches waiting list position, preference fields or the ordering logic at all.
Fair housing review is the most important approval here. Anything that changes how quickly different households move through a process is a fair housing question, so nothing prioritises and the measurements are defined so your own analysts can test for disparate effect.
The board adopted the admissions plan and will be asked publicly whether the list was automated. The answer needs to be a sentence: the order comes from the plan the board adopted, and nothing else touches it.
And resident leadership should see the unreachable measurement. It is the number that most directly concerns them and it belongs in front of them rather than only in front of staff.
One prior waiting list pull, measured retrospectively for how many removals for non-response followed a notification where every channel on file had already failed — with no access to position or ordering and no household contacted.
Retrospective, contact-only, and deliberately nowhere near the list order. It separates two things every authority currently reports as one: households who declined, and households who were never reached.
That separation tends to change what a board believes the waiting list is. It is a finding the authority keeps and can act on by changing how contact details are maintained, with no vendor involvement at all.
If it continues, the first grant covers contact maintenance between application and pull, with the ordering exclusion written into the scope document and the access list confirmed before anything runs.
Correct, which is why the access is absent rather than restricted — no credential reaches position, preference fields or the ordering logic at all, and that is checkable by inspecting the access list rather than reading a policy. The order comes from the admissions plan your board adopted in a public process. If a vendor offers to help you prioritise the list, that is the part to refuse regardless of how it is described.
Some did, and the whole point of the retrospective measurement is to stop reporting those two groups as one. A household who found housing and a household whose letter went to an address from three years ago are recorded identically today, and only one of them was served. If the measurement shows most non-responses followed channels that were still working, your process is fine and you should not buy anything.
Which is exactly why maintenance between application and pull is the band rather than a better letter at the end. The measurement to watch is reachability at the pull, not messages sent. And where nothing works, the design reports unreachable BEFORE a removal for non-response rather than after — so the decision to remove is made by your staff knowing the household was never reached, which is a materially different decision.
It is, and the handling defaults to the more protective reading wherever a record does not clearly say otherwise — because that is the failure direction that cannot be undone. Those protections are handled under their own rules rather than under a general privacy policy, and your counsel should confirm the specific handling before anything runs rather than after. If it cannot be met, the scope closes.