Large procurement departments track dozens of metrics. A smaller importer doesn't need dozens. Four, tracked consistently, catch most of the problems that actually matter before they become expensive.
In short: on-time delivery rate, defect or rejection rate, supplier concentration, and landed cost variance are four metrics a buyer without a dedicated procurement function can track with a simple spreadsheet, and each one exposes a different kind of risk that's easy to miss order by order but obvious once it's tracked over time.
On-time delivery rate
This is simply the share of orders that shipped within the agreed timeframe, tracked supplier by supplier over several orders rather than judged on any single one. A single late order can happen to a good supplier for a genuine reason. A pattern of lateness, visible only once it's tracked across multiple orders, is a different and more useful signal, and one that's easy to miss when each late order gets its own individual explanation in the moment.
Defect or rejection rate
Tracking the percentage of units failing inspection, or the rate of customer-reported defects if inspection data isn't available, over time and by supplier reveals patterns a single bad batch doesn't show on its own: whether quality is stable, improving, or quietly declining, the pattern behind quality fade. A rejection rate that creeps upward gradually across several orders is far more informative than any single order's result, and far easier to catch with a tracked number than with memory alone.
Supplier concentration
This tracks what share of total volume, by product category or overall, sits with a single supplier. High concentration isn't automatically bad; it often reflects a strong, well-earned relationship. But it's worth knowing the number deliberately rather than discovering it only when that supplier has a problem and there's no visibility into how exposed the business actually was. A simple rule of thumb: know the answer to "what happens if this supplier can't deliver next month" before it becomes a live question.
Landed cost variance
This tracks how much the actual landed cost of a product category deviates from what was planned or quoted, over time. A landed cost model that's consistently accurate is valuable on its own; one that's consistently off in the same direction, freight always running higher than estimated, for example, is a signal that the model itself needs updating, not that each individual shipment is an unlucky exception.
Why tracking beats reacting
None of these four metrics require sophisticated software. A simple spreadsheet updated after each order is enough to reveal patterns that are invisible order by order and obvious once plotted over several months. The value sits in the trend across data points, and in catching a shift early enough to act on it rather than discovering it retroactively after it's already cost real money.
Key takeaways
- On-time delivery rate reveals a pattern of lateness that a single late order doesn't show on its own.
- Defect or rejection rate, tracked over time and by supplier, is how quality fade actually gets caught before it's a serious problem.
- Supplier concentration should be a known, deliberate number, not something discovered only when a supplier fails to deliver.
- Landed cost variance flags whether a costing model needs updating, beyond whether a single shipment ran over.
- A simple spreadsheet tracked consistently outperforms memory and order-by-order judgment for catching all four patterns early.
Tracking these numbers across a portfolio of suppliers and product categories is part of what turns sourcing from a series of individual decisions into something closer to a managed process.
