Do not rank suppliers from incomparable inspection coverage

A ranking can make incomplete information look more decisive than it is. Before building a supplier score, inspect what each number counts and which work was never examined. A league table is not a neutral presentation when its rows have different evidence behind them.

This fictional team asks for a dashboard that ranks suppliers by recorded defects. Supplier North has 40 deliveries in the selected period, all inspected, with four deliveries containing a recorded defect. Supplier West has 40 deliveries too, but only five were inspected, with one recorded defect. The other 35 West deliveries are unexamined in the available system.

Keep three counts separate

The proposed dashboard initially compares four deliveries with recorded defects against one and places West ahead. Dividing by all deliveries would produce another apparently neat comparison, but it would still treat unexamined deliveries as though their condition were known. The data does not support that interpretation.

The team also examines why inspection coverage differs. In the fictional process, North's deliveries followed a complete inspection procedure, while West's inspected examples were selected after staff raised questions. The groups are not simply equal samples with different sizes. A rate calculated from the inspected subset would describe that subset, but it would not establish a fair overall supplier comparison.

The inspection preserves the underlying records. North's four affected inspection records remain four affected records. West's one remains one. The 35 unexamined deliveries do not become passed inspections, and an absent inspection result is not recoded as “no defect.” No ranking formula can repair the missing and differently selected evidence by itself.

Choose a view that supports the actual staff task

The purchasing owner needs to identify records requiring follow-up and understand the coverage of the evidence. The first release therefore shows delivery-level inspection status, recorded findings and the counts above. It provides a route to the actual inspected record and clearly separates unexamined work. It does not assign an overall performance position.

This is a bounded product choice, not a decision that supplier assessment is impossible. The organization can still make commercial judgments using its broader knowledge and approved procedures. The software avoids presenting an unsupported numerical conclusion as though it were established by the available records.

Verify the chosen result

The reviewer opens a North delivery with a finding and confirms that its inspection and finding remain visible. A North delivery with a completed inspection and no recorded defect shows that specific outcome. A West delivery without an inspection shows “Not inspected in this record,” rather than a green pass or a zero-defect result.

The totals reconcile to 40 deliveries for each supplier. West's five inspected and 35 unexamined records remain distinct. Adding an authorized inspection result to one previously unexamined fixture changes its coverage category and the relevant finding count; it does not silently update a hidden overall ranking. The selected period is visible so a later reader knows which deliveries the counts describe.

Define what would justify revisiting the score

The owner records the conditions for reconsideration: an agreed assessment purpose, comparable inspection selection and definitions, adequate evidence for that purpose, and a qualified review of the proposed interpretation. Those conditions are business and analytical decisions, not merely a request for more dashboard development.

Until then, the coverage view is the approved result. It answers a useful question—what has actually been examined and what needs attention—without claiming to answer a different one about overall supplier quality. Workflow application planning can help make that distinction visible before a polished score becomes difficult to question.