For finance ops leads
Finance ops: the recovery queue nobody staffed
Every finance organisation eventually receives a list of discrepancies. Heroic responses — someone spending a feverish week chasing the top items — produce stories, not systems. Six weeks later the list regrows and the hero has another job. What works instead is boringly mechanical: triage rules anyone applies identically, floors that pre-decide arguments, SLAs anchored to payout mathematics rather than politeness, and templates that make the right communication the lazy option.
Staffing the queue: roles and floors
One owner triaging daily at fixed time beats reactive whenever-ing. Floors do the pre-deciding: below $X absorb automatically (conversation cost exceeds recovery), above $Y escalate (materiality deserves seniority), between them standard playbook. Floors encode judgement once so individuals stop re-litigating values daily — the difference between a policy and a mood.
SLAs anchored to payout windows
Calendar-based SLAs ("resolve within 30 days") ignore the physics: recoverability decays per payout cycle. Anchor targets to the seller’s payout schedule instead — act before next payout where possible, escalate dormant-account items immediately to netting-or-write-off tracks. The queue’s urgency gradient should mirror money’s actual decay curve.
The escalation ladder
Standard ladder per finding: verify live state → attempt recovery per floor → netting offer if relationship warrants → support case if API path closes → write-off with documented EV. Skipping rungs wastes escalation capital; skipping write-offs breeds zombies. Ladder discipline is what makes the queue trustworthy enough to automate later.
Weekly metrics that matter
Four numbers weekly: findings opened/closed, dollars recovered, aging distribution, causes ranked. Trend beats absolute — improving closure velocity matters more than this week’s total. Causes-ranked feeds engineering directly; the metric review doubles as the roadmap meeting nobody schedules separately.
Templates as infrastructure, and FeeGuard’s role
Notice, netting and closure templates turn correct communication into copy-paste; ad-hoc emails are where tone disasters live. FeeGuard supplies the queue itself — states, floors-as-thresholds, evidence attached per finding, CSV exports matching your sheet columns — so the operating model above configures onto it rather than fighting it.
Common questions
How many hours does this cost weekly?
Modelled: triage minutes per finding once floors exist, plus exception time. The spreadsheet era’s cost was re-derivation, not resolution.
Beyond detection, what tooling do we need?
The queue, states, CSV export and audit trail cover the workflow. Most teams add nothing else for months.