Operated function · eligibility
A salaried applicant proves income with one document. Somebody with three irregular jobs, cash work, or a benefit that changes monthly cannot — and a process that treats both the same has not been neutral, it has quietly selected against the second.
The requirement list was written for a straightforward case. Proof of income, proof of address, proof of identity, proof of the qualifying circumstance. Each item has an expected document, and the expected document assumes a life with a payslip, a tenancy in your name, and a passport that has not expired.
Applicants who do not fit that shape submit something adjacent and are asked again. They usually cannot produce the expected document at all — not because they do not qualify, but because the evidence of qualifying looks different for them. Several give up at this point, and giving up is recorded as withdrawal rather than as a process failure.
On the staff side the work is comparison: does this document satisfy this requirement. That judgement is made differently by different people, because what counts as sufficient is largely custom rather than written — so two identical applications can receive different requests depending on who opened them.
Redetermination arrives on a cycle and repeats the whole thing for people whose circumstances have not changed, because nothing carries forward and the safest thing for an individual assessor is to ask again.
And the process metric is throughput or accuracy, neither of which notices that the population reaching approval is systematically narrower than the population entitled to it.
What counts as sufficient evidence for a requirement is held as individual custom rather than as a written standard with alternatives — so applicants who cannot produce the expected document depend on which assessor they get.
Writing it down is the whole intervention, and it is the part that is normally skipped because it appears to be documentation rather than work. For each requirement: what is being established, the expected evidence, and the acceptable alternatives — plural, deliberately, because the alternatives are what make the process reachable for the people it was written for.
Once sufficiency is written, delegation becomes safe, because the delegated act is comparing a document against a written standard rather than exercising discretion. It also becomes consistent: the same application receives the same request regardless of who opened it, which is a fairness property before it is an efficiency one.
The abandonment data is the second half. Recording which specific requirement an applicant stopped at turns a withdrawal number into a map of where the process is unreachable — and that is actionable, usually by adding an alternative rather than by chasing harder.
And redetermination becomes proportionate, because what was established and when is recorded, so an unchanged circumstance can be confirmed rather than re-evidenced from nothing.
The determination itself never moves. Whether somebody qualifies is a decision with legal and financial consequence for a person, and it belongs to your staff under your authority — always.
Consistency of what is asked for — measured by identical applications receiving identical requests, sampled across assessors before and after a written standard exists.
Where applicants stop — measured by abandonment by specific requirement, against a baseline where withdrawal was a single undifferentiated number.
Requests per application — measured by separate evidence requests sent before a file is complete.
Whether alternatives are actually offered — measured by requests naming acceptable alternatives, as a share of all requests.
Redetermination effort on unchanged circumstances — measured by items re-evidenced from nothing versus confirmed from a prior record.
Assessor time on judgement rather than comparison — measured by hours on determinations versus on document matching, sampled the same way before and after.
any eligibility determination, denial, appeal outcome or exercise of programme discretion, and no judgement about whether a document is genuine or an applicant is truthful. Nothing here decides who qualifies. It establishes what is required and whether what arrived meets a written standard — the determination is a decision about a person’s entitlement and it stays with your staff.
Cases stay in your eligibility or case-management system and documents land in your existing store. Nothing migrates and no second determination record exists, because two records of whether somebody qualifies is a due-process problem.
The eligibility rules and the sufficiency standard are yours: written by your programme staff, versioned, owned. They encode your statutory and policy position, and a standard that lived only in a vendor would be a governance dependency rather than an asset.
Where a verification source can be checked directly — an authoritative record you already have access to — that is used in preference to asking an applicant for a document, because the cheapest evidence request is the one that never has to be made.
What is collected is bounded by the requirement. A process that asks for a full bank statement to establish one income figure has taken far more than it needed, and in a programme setting that over-collection is both a privacy failure and a deterrent.
Sufficiency is checked; genuineness is not. Whether a document is authentic and whether an applicant is truthful are judgements with serious consequences, and they route to your staff. A system that flagged suspected fraud would be making an accusation, and it does not.
Every request and every receipt carries a record: what was asked, when, what arrived, what was still outstanding, and against which version of the standard it was matched. That is what answers an appeal.
Accessibility and language access are part of the requirement rather than an addition. An applicant who cannot read the request cannot satisfy it, and that failure would be recorded as their abandonment.
Operational access is not permission to train. Applicant circumstance data does not become material improving anything serving another organisation.
Legal owns the sufficiency standard, because it is a statement of what your programme accepts as evidence and it will be cited in an appeal. Writing it down is an improvement in your legal position, not a risk to it — the current unwritten version is the one that is hard to defend.
Whoever answers for equity of access should see the abandonment-by-requirement data first. It is the most useful artefact this operation produces and it is frequently uncomfortable.
Where an obligation attaches through statute, funding conditions or a data class, it is marked applicability-gated rather than presented as standing.
One programme over one trailing period — read-only, no applicant contacted — measuring abandonment by specific requirement and comparing requests issued across assessors on similar applications.
The observation phase produces the abandonment map: which specific requirement people stop at, and whether that differs by applicant circumstance. It contacts nobody and it is usually the most consequential thing the engagement produces.
The consistency comparison arrives with it. Where similar applications receive materially different requests depending on the assessor, that is a fairness finding your programme leadership can act on directly — with a written standard and no operated verification at all.
If you continue, the first delegation is matching submitted evidence against a written sufficiency standard on one requirement, with determinations untouched and alternatives named in every request.
The determination cannot and is not — it stays with your staff under your authority, and that is stated in the scope before anything is connected. What is delegated is comparing submitted evidence against a written standard your legal team owns. If your position is that even that comparison is a statutory act, the honest scope is the observation phase, and the abandonment map it produces is still the most useful thing here.
It will expose it, and the inconsistency already exists — it is simply undocumented, which makes it harder to defend rather than easier. In an appeal, an unwritten standard applied differently by different assessors is the weaker position. Writing it down is a legal improvement, and the discomfort is in the transition rather than in the outcome.
Where statute specifies a document, that is binding and no alternative exists — and the standard should say so plainly. In most programmes the statute specifies what must be established and the document list is administrative custom, and that distinction is worth establishing explicitly. The abandonment map shows what the current list costs in reachability; whether to change it is your programme leadership’s decision and possibly your legislature’s.
The current process is already sending them requests — the question is whether those requests are clear, name the acceptable alternatives, arrive in a language and format they can use, and happen once rather than three times. Where an applicant is flagged under your own vulnerability policy, the case routes to your staff rather than into a sequence. The abandonment data usually shows the existing process is what is failing this group.