AR Collections Forecast

Historical trends

Where the business has actually been, not just where it is today. Portfolio overdue % runs Jan–Jul 2026 from the collections team's own monthly tracker; segment and customer detail is only available for the three months with full aging detail (May–Jul).

Portfolio overdue %, Jan–Jul 2026

By segment, overdue % (May–Jul 2026)

The baseline

Forward billing forecast

PAPM recharge data only captures shared-services billing, not every AR line. Where a customer's matched PAPM figure looks like a small slice of their known current billing pace, their own "Future Due" balance is used as the monthly forecast instead — see Methodology.

Collections behaviour by segment

Real contracted payment terms (from the Payment Terms field) and a real, empirically-fit monthly collection rate — calibrated from the actual 31-May, 30-Jun and 28-Jul snapshots, not inferred from a single month.

Scenario forecast, Aug–Dec 2026

Sensitivity: what it takes to hit 20%

Dec-26 overdue % by monthly collection pace (how fast the overdue balance itself shrinks or grows, independent of billing) and monthly billing growth (which dilutes the ratio by growing the book). Bordered cell is the closest combination to the 20% target.

Top customers

The 25 largest overdue balances, with commentary from the collections team's own tracker where available.

Methodology & assumptions

What "overdue" means

Overdue % = (Total AR − "Future Due" bucket) / Total AR, evaluated at each month's close. This matches the collections team's own "Landing Overdue %" convention in the source workbook and ties out to 33.0% as of 28 Jul 2026. The same definition, applied to the 31-May and 30-Jun snapshots, gives 32.9% and 29.5% — the metric has moved in a fairly narrow band recently, not on a runaway trend in either direction.

Historical collections behaviour (May–Jul actuals)

Two more monthly snapshots (services - 31.05.26.xlsx, services - monthly 30.06.26.xlsx) give three real data points instead of one. Each customer's real contracted payment terms come from column K ("Payment Terms") on the DATA tab of services - 28.07.26.xlsx — the current month's own invoice-level terms (30/45/60/90 days from invoice or EOM, Immediate, etc.), $-weighted per customer — portfolio median 30 days, but Controlled-segment relationships often run 45–60+. From terms we derive p, the share of a customer's "not yet due" balance that ages into overdue each month (30-day terms → p=1; 90-day terms → p=1/3). Comparing real Future-Due and overdue balances month to month (May, June, July), with p fixed from those real terms, isolates a real monthly collection rate for each of the two live transitions (May→Jun, Jun→Jul) — blended $-weighted, recent transition counted double. Segment rates this produces: Controlled and 3rd Party collect fast (~80–85%/month), JV moderate (~50%), Connected slow (~8–17%) — consistent with Connected's much higher overdue share in the baseline. Customers with AR ≥ $250k get their own rate; everyone else uses their segment's blend. See "Collections behaviour by segment" below to drill into which customers sit in each segment.

Forward billing

PAPM's ICSB Flow Map lists 32,694 initiator→receiver recharge lines; 12,146 are explicitly tagged /MONTH and 245 /YEAR (converted ÷12) — these are treated as recurring. ~20,300 untagged lines (recruitment fees, one-time project/setup charges, ad-hoc services) are excluded. Pluto Flow Map (external, non-Oracle customers) has real invoice dates Jan–May 2026; each customer's forecast is their 5-month average. Matching PAPM receiver names to AR customer names goes through four passes: exact match; a customer's full name as a prefix of a PAPM row (catches brand/BU sub-lines under one legal entity); a conservative fuzzy match; and a same-root-word consolidation for divisional splits of one company (e.g. "ALLIED ENTERPRISES LLC" picks up both "ALLIED - RETAIL" and "ALLIED - SELECTIVE"), guarded so it never merges two genuinely different companies that happen to share a surname (the several distinct "CHALHOUB..." entities, for instance) or a brand-named sub-account into its parent. Near-duplicate spellings within PAPM itself (e.g. a stray "...WLL" vs "...LLC" of the same receiver) are consolidated first. This lifted direct PAPM coverage from an initial 120 customers / 76% of AR $ to —. Where a matched customer's PAPM figure is still under half their current Future-Due balance, their Future-Due balance is used instead.

Forecast mechanics

An initial month-by-month simulation (aging pool + the empirical rate above, compounded five times) was tested and rejected: with only two real transitions to fit a rate from, compounding it five months out extrapolated a pace of change well beyond anything the real data has ever shown — particularly for the small, slow-paying Connected segment, where estimation noise gets amplified heavily by 1/rate. Instead, each scenario's overdue % is built as a direct, bounded drift off the 33.0% July actual, paced at a maximum of 3.5 points per month in either direction — the largest real month-to-month swing across May, June and July. The segment collection-rate hierarchy above sets each scenario's direction and relative speed, not a compounded multi-period formula. Total AR (the denominator) still rolls forward from real billing and aging dynamics.

Scenarios

Realistic — overdue % held flat at 33.0%: no assumed change in collections behaviour. AR grows with the billing forecast; the overdue balance grows proportionally, so the ratio doesn't move.
Downside — overdue % drifts up ~1.2 points/month (behaviour erodes, billing ramps further toward the PAPM-implied run-rate, a summer collections slowdown).
Upside — overdue % improves ~2.0 points/month (a managed collections push, plus faster legacy-debt recovery).

Known limitations

Only three real monthly snapshots exist, giving two real transitions to fit behaviour from — real, but a small sample; segment-level rates (especially Connected's) carry real estimation uncertainty.
PAPM billing coverage is now strong (93.5% of AR $ has a direct PAPM match) but not complete; the remainder falls back to a Future-Due-based proxy.
Scenario pacing is capped at the fastest month-to-month move the real data has shown — a deliberate, disclosed guardrail against over-extrapolating from two data points, not a claim that faster or slower months are impossible.
The model doesn't distinguish credit notes/disputes from genuine slow-pay; large negative balances (credit positions) are floored at zero when rolled forward.
Portfolio-level trend goes back to Jan-26 (the collections team's own tracker); segment- and customer-level trend only goes back to May-26, the earliest month with full aging detail on hand — Jan-Apr only exist as portfolio totals in the source files.