How to spot supplier delivery problems using your PO history

Decide whether you are measuring order lines, shipments or whole orders. A purchase order with ten lines can look successful when nine arrive on time, even if the missing line stops production. For a first analysis, order lines are often easier to explain because the required quantity and commitment belong to the same item.

Keep the original promised date and subsequent revisions. Measuring only against the last revised date can make a slipping supplier appear reliable. You may want two measures: performance against the original commitment and performance against the latest commitment accepted by your team.

For the example here, an order line is on time when the full required quantity is received by its original promised date, using calendar days and no grace period. Your team should choose and document its own tolerance, receipt definition and treatment of partial shipments before comparing suppliers.

Export the supplier ID, PO and line number, part, order date, original promised date, accepted date revisions, required quantity and receipt history. Include the source system and a link or reference to the source line. Keep currency and spend if you intend to rank financial exposure.

Do not fill missing receipt dates with today’s date. A blank can mean an open order, a missing integration, or a receipt that has not been posted. Separate open commitments from completed deliveries and put incomplete records into a review list. Otherwise, you can turn missing data into invented delay.

Northfield buys castings from Acme. Six completed order lines show the following delivery history. The downloadable CSV contains these same rows, with the fields used in the calculation.

For a completed line, days late = max(0, full-receipt date − original promised date), measured in calendar days. A delivery two days early contributes zero days late, not minus two. Otherwise, early arrivals could cancel the delays you are trying to understand. Track days early separately if early deliveries create storage or scheduling problems.

Consider two additional fictional lines, separate from the six-row CSV. PO-201 requires 100 units by August 20 and receives all 100 on August 18: it is on time, with zero days late. PO-202 also requires 100 units by August 20; 80 arrive on August 19 and the remaining 20 arrive on August 24. Under the full-receipt policy, PO-202 is four days late. The first receipt does not complete the line.

At an August 22 cutoff, PO-202 would instead be an open overdue line: 20 units remain outstanding, two days past commitment. Its final delivery lateness is still unknown. Keep that open-line view separate from the completed-line average, and record any agreed change to the required quantity before deciding that a line is complete.

A supplier of custom tooling and a supplier of stocked fasteners have different commitments. Compare the same supplier over time, then compare similar parts, sites and order types. Report the number of eligible lines beside each percentage so that one late line out of two does not look like fifty late lines out of a hundred.

Look at delivery lateness alongside actual lead time, promise revisions and outstanding commitments. A rise in actual lead time can matter even while shipments remain on time, because your team may be accommodating the change by placing orders earlier. Check whether a new part mix or your own late approvals explain the change.

Use a recent window and a longer baseline with clearly stated dates. If a line is open and overdue, describe it as an overdue commitment at the analysis cutoff. It belongs in today’s follow-up list, but it does not yet have a final delivery lateness.

A plant-closure article adds context only if it concerns the site making your part. A shared company name is not enough. Check the legal entity, location, event date and relevant part or business unit. An article about Acme Packaging in another country should not become evidence against Acme Castings.

Keep the source and what it actually says. Distinguish a reported closure from an inference about your orders, and retain evidence that contradicts the concern. A supplier’s confirmation that your work moved to an unaffected site changes the assessment; it should not disappear in a notes field nobody reads.

The Acme example supports a specific follow-up: four recent deliveries slipped by two, three, five and seven days. Ask what changed, whether open commitments are affected, and which recovery date the supplier can support. Attach the affected PO lines so both teams can discuss the same facts.

Prioritize the conversation by operational exposure. A small late order for a sole-source production part may deserve attention before a larger order covered by stock and an approved substitute. Keep delivery performance, inventory cover, part criticality and supplier response visible as separate considerations.

Semogram connects the data you already have, reconciles records that refer to the same supplier, and supports answers with evidence. The implementation starts with your source data and a defined supplier, order and receipt model. Publish a query for the bounded question you want to answer, then inspect the records and runs behind its output.

Ask your assistant to compare recent commitments with delivery outcomes and explain which suppliers need review. Treat its output as something to check: confirm the supplier match, window, denominator and evidence links before using it in a supplier discussion. If you add forecasting, evaluate it separately on your own history; a delivery trend is not itself a validated probability of future failure.