Deciding when a signal becomes an action: severity, repetition, batch and use judged together

Four factors together turn a signal into an action: severity, repetition, concentration and cause. Severity asks how serious the harm or risk is: a burn, a swallowed part or a fire risk weighs more than a cosmetic fault, even when reported once. Repetition asks how often similar reports arrive: several similar reports weigh more than one, especially when they arrive close together. Concentration asks whether reports share a batch, a part, a supplier or a use: concentrated reports point to a specific cause, while scattered reports may reflect general wear. Cause asks whether a plausible mechanism links the reports: a shared heating element, a loose fixing or a changed material gives the signal a story that scattered coincidence lacks.

A decision meeting should work through the factors explicitly, with the signal evidence in front of the group: the grouped reports, the batches and parts involved, the severity of the worst case, and any supplier or test information available. Each factor should be discussed and noted, including factors that argue against action, such as a plausible alternative explanation or a lack of concentration, since a decision that only records supporting factors looks one sided and will not withstand later scrutiny. The meeting should end with a decision, a named owner, and a date, even when the decision is to watch, since a watch decision without a review date is indistinguishable from inaction.

Action does not always mean recall. Available actions form a ladder: investigate further with testing or supplier enquiry, pause sales of an affected batch while investigating, update instructions or warnings, change a supplier or component, repair or replace affected units, or, where the signal warrants it, carry out a corrective action or recall. Choosing the lowest action that addresses the risk, while keeping higher actions available if investigation confirms a wider problem, is proportionate and preserves customer trust, since customers respond better to a seller who acts early and proportionately than to a seller who either ignores a signal or overreacts to every minor report.

Illustration: a seller of desk lamps received 30 reports of flickering from one batch, investigated with the supplier, and found a loose connection in that batch. The seller paused that batch, repaired affected lamps, and updated its incoming inspection for lamp connections, without recalling other batches. The report count is illustrative, and the lesson is to match the action to the batch and cause, rather than applying the same action to every product that shares a name.

Watching is right when a signal is weak on several factors: reports are few, minor, scattered across batches, and lack a clear cause, yet are similar enough to merit attention. Watching should be active, not passive: it means logging new reports against the signal, setting a review date, and defining in advance what new evidence would trigger action, such as a severe report, a batch concentration, or a repetition threshold for that signal. A watch decision recorded with those triggers is a decision; a watch decision recorded only as monitor is a postponement that may be forgotten.

Watching is wrong when severity is high, even if repetition is low, since a severe harm reported once may recur before a pattern forms. It is also wrong when concentration and cause are clear, even if reports are few, since waiting for more reports means waiting for more customers to experience the same fault. In those cases, investigation or precautionary action, such as pausing an affected batch, is the proportionate response, and the decision note should say so plainly, so a later reviewer understands why action was taken on limited numbers.

Record every signal decision, including watch decisions, with the signal summary, the four factors, the decision, the owner and the review or completion date. That record protects the seller when a decision is questioned later, by an authority, a retailer or a customer, since it shows the decision was reasoned from evidence at the time, not made casually or after the event. It also helps future teams, who can learn from past decisions without repeating the same debate for a similar signal, and it creates a decision history that improves with each case.

For the next signal, use a one page template: signal summary, severity, repetition, concentration, cause, decision, owner, review date. Complete it in the decision meeting, file it with the monitoring log, and review it on its review date. A template used consistently turns signal decisions from stressful improvisations into a routine that a team can perform calmly, even when the signal is serious and the pressure to decide quickly is high.