Risk Band

A risk band groups customers into a small number of ordered tiers based on a scorecard result, a bureau grade, or a blend of both. The purpose is not classification for its own sake: a band exists so that policy can differ across it. Approval authority, maximum limit, payment terms, deposit or security requirements, review frequency, and collection intensity should each be defined per band, written into the credit policy, and applied without renegotiation on every file. If two bands receive identical treatment, one of them is decoration.

Three or four bands is the practical range for most mid-market portfolios. Fewer than three cannot separate the routine from the watchful; more than five produces boundaries the team cannot explain and customers cluster at the cutoffs anyway. Set the cutoffs from your own loss experience rather than from the score vendor's suggested ranges, because the same bureau grade carries different loss rates in construction supply than it does in food distribution. Record the reasoning for each cutoff — it is the first thing an auditor or a new credit manager will ask about.

The most common failure is band drift: customers are banded at onboarding and never re-banded, so the portfolio's stated risk profile slowly detaches from reality. Bands should be recomputed on a schedule and on trigger events — a bureau score change, a payment-behavior deterioration, new financial statements — and a change of band should be an event that something acts on, not a field that quietly updates. The second failure is the permanent exception: an account manually held in a better band than its score supports, with no expiry and no recorded reason.