Section 1 — Client portfolio
Section 2 — Team strength
Section 3 — Reviewer portfolio
| Reviewer | Assigned | Signed off | Pending | Changes req. | On time | Min / close | Workload | Status |
|---|---|---|---|---|---|---|---|---|
| JPJordan ParkerPartner · CPA · 9 yrs | 16 | 13 | 3 | 0 | 100% | — | 42% | Active |
| MRMaya RuizReviewer · CPA · 6 yrs | 14 | 12 | 1 | 1 | 98% | — | 37% | Active |
| ECEllen CalderPartner · CPA · 21 yrs | 5 | 5 | 0 | 0 | 100% | — | 13% | Active |
| MFMarcus FinchPartner · CPA · 18 yrs | 3 | 2 | 1 | 0 | 100% | — | 8% | Active |
| TATom AlvarezStaff · no credential | 0 | 0 | 0 | 0 | — | — | 0% | Cannot sign |
Min / close is empty on every row, and that is the finding. `Review` has no timing fields, so the one number that says whether Bursar’s economics work — whether a reviewer’s minutes per close fall as the engine learns — is not captured anywhere. A reviewer whose time per close stays flat is a services business with a software price.
Section 4 — Pipeline intelligence
| Stage | Closes here | Txns | Roster past this | Avg age (d) | Blocked | |
|---|---|---|---|---|---|---|
| IngestTransactions and documents flow in | 0 | — | 100% | — | 0 | Healthy |
| CodeEvery line categorised to an account | 2 | 1,284 | 95% | 1.2 | 0 | Healthy |
| ChaseMissing receipts and statements requested | 3 | — | 92% | 4.1 | 1 | Blocked |
| ReconcileBooks proven against the bank statement | 4 | — | 82% | 2.0 | 0 | Watch |
| ReviewA credentialled reviewer verifies and signs | 5 | — | 68% | 0.8 | 0 | Watch |
| DeliverPackage shipped on the firm’s letterhead | 24 | — | 63% | — | 0 | Healthy |
Adjacent work this table grows to hold
Section 5 — Revenue & billing
Section 6 — SLA clock & connection expiry
Re-authorisation queue — the chase that needs the client, not the firm
| Client | What lapses | In | If it lapses | |
|---|---|---|---|---|
| Pier 9 Seafood | Bank feed authorisationChase · open banking | 2d ago | Feed already stale — no new lines since 26 Jun | Broken |
| Lakeside Fitness | Bank feed authorisationWells Fargo · open banking | 9d | Coding stops when it lapses | Expiring |
| Northside Vet Clinic | QuickBooks refresh tokenIntuit · 100-day idle | 21d | Re-consent needed; 5-year hard cap applies too | Expiring |
This table is the analogue of the source dashboard’s DSC expiring column, and it is the most under-served problem in the category. QuickBooks refresh tokens die after 100 idle days with a five-year hard ceiling, and open-banking authorisation lapses roughly every 90 days — and re-authorising needs the client’s credentials and MFA device, not the firm’s and not ours. It fails silently, so the first symptom is a close that will not reconcile. Every document-chase product on the market ignores it.
Section 7 — Client risk
What the score is built from
Clients above routine
| Client | Score | Band | What drives it |
|---|---|---|---|
| Pier 9 Seafood | 7.4 | Critical | Blocker flag · statement outstanding day 6 · bank feed lapsed |
| Lakeside Fitness | 3.1 | Action required | Awaiting sign-off · feed expires in 9 days |
| Northside Vet Clinic | 2.8 | Action required | Awaiting sign-off · token expires in 21 days |
| Birch Street Dental | 1.4 | Watch | One receipt parked (23/24 documents) |
The weights are published on purpose. The source dashboard has a CLIENT_RISK_ENGINE column whose number arrives with no derivation, and an unexplained risk score is one a partner cannot act on — they can only agree or disagree with it. Every factor here maps to a column we already store.
Section 8 — Executive indicators
The last two cells are the ones to watch. Automation rate and reviewer minutes per close are not features — they are the instruments that decide whether the unit economics hold, and both are missing. If a “90% automated” product is really 60% automated, the gap is human hours that scale linearly with customers, and margin inverts silently while growth makes it worse. That is the documented failure mode of the two best-funded companies in this category.
Section 9 — Integrity watch
| Check | How it is detected | State | |
|---|---|---|---|
| Prior-year Retained Earnings moved | Snapshot at delivery, diff each period | Not implemented — feature-map B7 | Not built |
| Reconciliation was forced | Memo/Description = “Reconcile Adjustment” | Not implemented — feature-map B9 | Not built |
| Reconciliation abandoned, not plugged | Unreconciled period on a bank account | Not implemented — the more common case | Not built |
| Period never locked after delivery | Bursar does not record a period lock yet | Not modelled | Not modelled |
This section has no counterpart in the source dashboard; it exists because a delivered close can be falsified afterwards and nothing currently notices. QuickBooks rolls net income into Retained Earnings as a computed, non-posting figure — it has no register and is recalculated on every report run — so editing a transaction in a closed year moves it with no journal entry to explain it. Snapshotting prior-year Retained Earnings at delivery and diffing it each period catches the whole class in a single comparison.
Nine sections, one screen, no pagination — the density is the design. Adapted from an executive practice-management dashboard for an Indian CA firm; its structure transferred, none of its content did. Every service line, currency, role and fiscal year in the original is India-specific, and the mapping onto Bursar’s domain is documented in src/data/birdsEye.ts.