Early Warning Signal

An early warning signal is any observable that tends to move before default does. They fall into four families. Behavioral signals come from your own ledger: days beyond terms stretching, a shift from checks to the last day of the grace period, part-payments appearing, promises to pay broken, disputes raised late in the aging cycle as a stalling device. Financial signals come from statements: gross margin compression, operating cash flow diverging from net income, rising days inventory outstanding, leverage climbing against tangible net worth. External signals come from bureaus and public records: score declines, new liens or judgments, UCC filings by a lender, suits. Relationship signals come from the commercial team: a delayed statement, a change of CFO, a key contract lost, a facility quietly closed.

Signals trade lead time against reliability, and it is worth being explicit about which you are buying. Financial-statement deterioration gives the longest warning but arrives late and infrequently — often eight months after the period it describes. Behavioral signals from your own AR are the timeliest and cheapest, because you own the data and it updates daily, but individually they are noisy: one slow payment is an accounts-payable run, not a crisis. Public-record signals are the most specific and the closest to the end of the story.

The practical rule is that composites beat singles. A customer stretching from 34 to 51 days is a question; a customer stretching to 51 days while its bureau score falls two grades and a supplier files a UCC is an answer. Formal composite tools such as the Z-Score encode exactly this logic for financial data, and the same principle applies to blending behavioral with external signals. Define in advance which combinations trigger a review, so the response is a policy rather than whoever happened to notice.