A cold chain KPI is a number tracked over time to judge whether a shipping or storage operation is actually working, not just running. The useful ones measure an outcome, did the product arrive in range, was it usable on arrival, not an activity, how many shipments moved or how many sensors got deployed. A network can look busy on every activity measure and still be losing product quietly if nobody is tracking the outcomes those activities were supposed to produce.
Five numbers cover most of what matters: excursion rate, on-time-in-full at temperature, product loss, logger return rate, and cost per shipment. None of them means much read alone; together they show whether a network is actually protecting product or just moving it.
Excursion rate
Excursion rate is the share of shipments that recorded a temperature excursion at any point in transit or storage, regardless of whether the product was ultimately judged usable. It is the earliest warning number in the set, since a rising excursion rate on a specific lane or facility usually shows up well before a loss or a customer complaint does. Read it by lane and by season, not as one network-wide figure, because a single average hides the one route or the one summer month actually driving the number up.
On-time-in-full at temperature
Ordinary on-time-in-full only checks whether a shipment arrived complete and on schedule. Adding the temperature condition turns it into the number that actually matters for cold chain: a shipment that arrived on time and complete but outside its qualified range is not a success by any real measure, and a KPI that does not check temperature will call it one anyway. Tracking the temperature-adjusted version alongside the ordinary one exposes exactly how much of a network's apparent on-time performance is really performance and how much is a shipment that happened to arrive warm on schedule.
Product loss
Product loss ties the operational numbers back to money and, in the cases that matter most, patient or consumer safety: units actually written off because they left their qualified range for longer than the product could tolerate. It lags the other measures, since a loss is usually confirmed well after the excursion that caused it, which is why it works best as a check against the leading numbers rather than the first thing a team watches day to day. A rising excursion rate that is not yet showing up in loss is either a network catching problems before product is affected, or a loss about to surface with a delay.
Logger return rate and cost per shipment
Logger return rate, the share of temperature data loggers that actually make it back for reuse or readout, is a quieter number that still drives real cost: a logger not returned is a logger bought again, and a chronically low return rate on a specific lane or customer usually points at a process gap rather than genuine loss. Cost per shipment closes the set, the full landed cost of keeping one shipment in range, packaging, coolant, logger, tracker fees and any premium freight, tracked so a network can weigh whether a proposed upgrade actually pays for itself against the loss and excursion rates it is meant to improve.
Trends over time, not a single snapshot
A single month's excursion rate or loss figure means little on its own, since normal variation between lanes and seasons can move a number up or down without anything having actually changed in how the network is run. The useful read is a trend against the same lane's own history and the same season a year earlier, not a comparison against an unrelated route or a different time of year. A lane that runs consistently worse than its own past record, even while staying within a target range, is showing a real change worth investigating before it turns into a loss.
Roles that actually use the numbers
Quality managers own excursion rate and product loss, since both tie directly to whether product reaching a patient or a customer was actually fit to use. Logistics and network managers own on-time-in-full at temperature and cost per shipment, the two numbers that describe whether the network runs efficiently as well as safely. Reporting all five on one shared dashboard, rather than splitting them across separate departmental reports that never get compared side by side, is what lets a business actually see the trade-off between running cheaper and running safer, instead of each side working from its own partial picture.
Measuring outcomes, not activity
The common failure across all five is substituting an activity count for an outcome: reporting how many shipments moved instead of how many arrived in range, or how many sensors got fitted instead of how many loggers actually came back. Activity numbers are easy to collect and almost always trend upward, which makes them comfortable to report and useless for deciding whether anything is actually getting better. A cold storage network with five honestly tracked outcome numbers, even ones that look uncomfortable some months, is being run more carefully than one reporting only volume moved.