Tracking whether a recall actually worked: response rates, stragglers and second waves

Messages sent is a vanity count. It grows every time a notice goes out, whether or not anyone acts. Safe units is the honest count: units returned, repaired and verified, replaced, or confirmed scrapped by the buyer where that confirmation is credible. A recall that contacted everyone but made few units safe has communicated well and protected poorly, and the tally should show that plainly.

The tally needs a denominator too: the total affected units, by channel and by country where the data allows. A response rate without a denominator is a percentage of nothing. Where the denominator is uncertain, the tally should state the basis, such as units shipped in the affected batch, and update it when better figures arrive.

Response typically surges in the first days after a public notice, then slows into a long tail. That shape is normal, not a sign of failure, but the tail deserves attention, because the units in the tail stay in use the longest. Plotting responses by week, rather than reading a single total, shows whether the tail is still moving or has flatlined, and a flatline is the signal to change tactics.

The curve should be split by channel. Direct buyers often respond fastest, marketplace buyers slower, and retail buyers slowest of all, because retail buyers may never see a notice unless it reaches the shop or the media they follow. A single blended rate hides a channel that has barely moved, and a channel specific view tells the team where the next reminder should go.

Non response has causes, and causes have remedies. Some buyers moved house, so postal notices miss them. Some bought as gifts, so the holder never saw the seller's message. Some distrust messages about products they believe are fine, especially where the fault is invisible. Some simply never open marketplace messages. Each cause points at a different second route, from a public notice to a retailer poster to a message through a different channel.

Where the reason is known, it should be logged. A pattern, such as many undeliverable addresses from one retailer period, points to a data problem the seller can fix. Illustration: a seller of baby monitors finds low response from one retail chain's customers and learns the chain's till receipts carried no product code, so the seller places posters in that chain's stores instead of relying on purchase records.

A second reminder that repeats the first notice word for word adds little. A better reminder changes something: the channel, the wording, or the ease of action. A shorter message with a single action, a photograph of the product in a home setting, or a prepaid label enclosed rather than offered, can move buyers who ignored the first attempt. The reminder should also acknowledge that the buyer may have seen the first notice, rather than pretending it is new.

Timing matters. A reminder sent the day after the first notice will be seen as noise, while a reminder sent after the first surge has faded gives buyers a genuine second chance. The team should also measure the reminder's effect separately, so it learns which second route actually moved the tail and which merely cost postage.

A recall rarely moves evenly across countries or buyer groups. Language, media habits, retail structure and trust in official messages all vary, and a single average hides a country where response has stalled. Splitting the tally by country and by channel lets the team direct effort where it is needed, such as a translated notice in one market or a retailer campaign in another.

Buyer groups matter too. Trade buyers, such as hotels or schools holding many units, need a different approach from households holding one, because a single trade contact can resolve dozens of units at once. The tally should show trade units separately, so a handful of trade resolutions is not mistaken for broad household progress, or vice versa.

Closure is a judgement, not a percentage threshold this guide can set. The team should weigh the response achieved, the risk of the remaining units, the effort already spent on reminders, and the view of authorities involved. A recall for a low severity fault with a high response may close sooner than a recall for a serious fault with a middling response, where further effort is still justified.

Whatever the decision, it should be recorded with reasons, the final tally, and the plan for late returns, because buyers will still trickle in after closure. A closed recall that still accepts late returns gracefully, and logs them, protects the remaining holders without keeping the whole operation open indefinitely. That combination, a recorded decision plus an open door for latecomers, is the honest end of a tracking job.