# Institutional acceptance and evidence checklist

DRAFT PLANNING TEMPLATE — not an executed agreement, completed test, certification or delivery record. Prepared 19 September 2026. Complete, review and approve before institutional rollout. Do not enter real student data in public copies.

Record the deployment URL/build, test date, synthetic dataset, tester, expected result, actual result, screenshot/file and reviewer for each case. Leave untested items unapproved.

| Case | Acceptance evidence | Result / owner |
| --- | --- | --- |
| Administrator authentication | Login, logout, expired session, unauthenticated API denial | Pending |
| Student isolation | Student A cannot read or modify student B records | Pending |
| University isolation | Institution A cannot read/export institution B records | Pending |
| Staff roles | Approved administrator/counsellor/department matrix; allowed and denied actions | Pending: role model requires confirmation |
| Bulk onboarding | Valid CSV, invalid email, duplicates, malicious cells, delivery failure, seat limit, retry | Pending |
| Branding | Agreed logo/colours and approved-template enforcement demonstrated | Pending |
| Usage reporting | Event definition and count reconciled against synthetic source events; opt-outs respected | Pending |
| Document reporting | Drafts, versions, successful exports and unique documents distinguished | Pending |
| Skills reports | Validated framework, scoring rationale, limitations and human review | Pending: capability not established |
| Interview tools | Input, generated guide, written feedback, provider failure and cost limits | Pending |
| Private vault | Unlock, wrong passphrase, encrypt, lock, export, restore, lost-key warning | Pending |
| Institutional export | Fields, suppression, CSV safety and authorised access | Pending |
| Exit | Student continuity, export, provider deletion and backup schedule | Pending |
| Consent | Fresh browser has no optional requests; decline persists; selective consent and withdrawal | Pending |
| Accessibility | Keyboard, focus, labels, contrast and mobile readability | Pending |

A screenshot proves only its displayed state. Pair it with reproducible test steps and timestamps. Public sample CSVs are synthetic and cannot establish production usage.

Impact evidence: state period, timezone, unique-user rule, verification method, deduplication, test-account exclusion, opt-out coverage and reviewer. Do not sum per-feature users. Interview/employment outcomes require separate consented follow-up and denominator.

Partnership register: legal institution name, relationship type (executed/pilot/outreach/proposed/affiliate), authorised signatory, execution/expiry dates, scope, permission to name/logo, supporting document and reference contact. No entries are pre-certified.