Where the loss sits

Select a segment to see its ageing profile, then a bucket for the workings behind its loss rate.

Scenarios and the non-linearity demonstration

Each scenario's expected credit loss, and the probability-weighted result against the result computed once at weighted inputs. That comparison is the evidence IFRS 9 B5.5.42 requires on whether one scenario would suffice.

Data quality — ingestion diagnostics

3 checks raised warnings.

Retrospective review — what last period got wrong DB-RETRO-01

What last period’s estimate got wrong, and whether the current estimate’s assumptions have been confronted with that.

Movement decomposition

Assumption back-test

Prior ECL of 3600.00 was below the 4100.00 realised by 500.00, a optimistic lean. One period is an indicator, not proof; bias is a pattern across periods [ISA 540.14].

Segmentation — does it distinguish credit risk? DB-SEG-01

Whether exposures have been grouped on shared credit risk characteristics, or whether the data shows loss behaviour the segmentation does not distinguish. Client basis: Sales channel (retail / wholesale).

The segments' loss rates span 0.02607173 to 0.14218367, and every adjacent pair differs by at least the 0.01000000 tolerance. The segmentation distinguishes loss behaviour the data shows.

Data reliability — the estimate’s foundation DB-DATA-01

Whether the client data ingested is sufficiently complete and reliable to support the conclusion being drawn from it.

Provenance. source Client aged debtor ledger · via auditor upload · 10 rows as at 2025-12-31 · hash 8f4b276b64c9bc01

3 checks raised warnings. The full diagnostic results are in the data-quality panel above.

Subsequent receipts — the collection evidence DB-SUBSEQ-01

What the post-period-end collection experience says about the loss rates applied, and whether the estimate is consistent with it. Cash received to 2026-03-31 — 90 days after the reporting date — matched against the receivables provided on [ISA 540.21; ISA 560].

Collection by ageing bucket, against the applied loss rate

The comparison the block exists for: what the estimate provided for each bucket, and what the subsequent cash shows. Presented, not adjudicated — a balance unpaid at the cut-off may yet be collected.

Collection by segment

Exceptions

Post-date credit notes and receipts matching no period-end receivable. Each bears on the completeness or the validity of the population provided on [ISA 500.7].

• limitation — 1 receivable(s) were reduced by a post-period-end credit note. A credit note reverses the sale rather than collecting it, so these are surfaced as exceptions and excluded from cash collected.
• information — 1 post-period-end receipt(s) matched no receivable on the period-end ledger. This can indicate a customer omitted from the ageing or cash applied against an already-cleared balance, and bears on completeness of the population [ISA 500.7].

Matters affecting the estimate

Conditions the engine met in the data that bear on how far the estimate can be relied on.

• information — 2 observation(s) of the d61_90 → d90_plus transition in segment 'retail' exceeded 100% and were capped. A bucket that grows between periods reflects part payments or new charges rather than a loss rate above the exposure.
• information — 2 observation(s) of the d90_plus → write_off transition in segment 'retail' exceeded 100% and were capped. A bucket that grows between periods reflects part payments or new charges rather than a loss rate above the exposure.
• information — 2 observation(s) of the d90_plus → write_off transition in segment 'wholesale' exceeded 100% and were capped. A bucket that grows between periods reflects part payments or new charges rather than a loss rate above the exposure.
• limitation — 3 bucket-scenario combination(s) produced a loss rate above 100% before bounding, and were held at the exposure. An expected credit loss cannot exceed the amount owed, so the excess is not recognised — but a forward-looking factor pushing an already-total loss rate higher indicates the factor is being applied to buckets where it has no room to act, and the segmentation or the factor itself should be revisited.

Individual assessment — do the cash flows support the provision? DB-INDIV-01

Whether the discounted expected cash flows support the specific provision on this exposure, or whether the recoverable amount the evidence supports differs from the client's. Expected recoveries are discounted at each exposure's original EIR to 2025-12-31.

The auditor's range — how does the provision sit against it? DB-RANGE-01

Whether the range is narrow enough to evaluate the client's point estimate, and where the client's provision sits against it.

Client provision £300,000.00 marked against the band.

Proposed understatement of £20,000.00 — below performance materiality, raised for disposition (§17.4).

Auditor's range [320000.00, 450000.00] built from the primary estimate (366942.15), the vintage_loss_rate challenger (410000.00) and 2 scenario point(s). Point estimate is the primary run. Client provision 300000.00 is below range.

Overlay challenge — does each adjustment survive its own evidence? DB-OVERLAY-01

Whether this overlay addresses a risk genuinely absent from the model, whether its quantification is evidenced rather than asserted, and whether its governance is intact. 2 overlay(s): £275,000.00 increasing, £0.00 decreasing, net £275,000.00 — all overlays increase ECL.

Contradictions — where the evidence disagrees with itself DB-CONTRA-01

How the divergence between what the client states, what the client asserts and what the data shows is resolved. 2 of 2 evidenced element(s) contradicted; 0 agreed across layers; 0 single-sourced. A contradiction cannot be deleted — only dispositioned with a rationale.

Open items (§10.4): Collateral policy — no document, inquiry response or analytic evidences these yet.