OTIF: calculate delivery performance from the lines due
Calculate OTIF with eight dated delivery obligations. Keep unreceived lines in the denominator and track recovery without rewriting the original result.
OTIF means on time, in full. It measures whether the required goods arrive within the agreed delivery window and in the required quantity. A manufacturer can use it to review supplier deliveries or its own deliveries to customers, provided each measure names its receiving location, date and counting unit.
The calculation needs the obligations that were due, including the ones that did not arrive. This worked example follows eight purchase-order delivery obligations from their original dates to receipt evidence and recovery actions. It is a practical method you can reproduce in a spreadsheet.
Define the measure before calculating it
McKinsey and the Trading Partner Alliance document different OTIF definitions and propose a case-quantity measure for consumer goods. The binary line method below is a different, explicitly chosen convention. Use the definition agreed for your customer or supplier program when reporting contractual performance.
For this internal supplier review, write down these rules:
| Rule | Convention used here |
|---|---|
| Counting unit | One original order-line delivery obligation with an agreed quantity and date; one stable ID |
| Baseline | First supplier-confirmed arrival date at our receiving site |
| Eligible period | All obligations with that original date from November 2 through November 6, 2026 |
| Deadline | 18:00 receiving-site local time on each due date; earlier arrival is allowed in this example |
| Success | The full required quantity of the correct part and revision has arrived at the right site by its deadline |
| Observation cutoff | November 6, 2026, at 18:00 in the same local time zone |
An order with two delivery dates agreed in advance can have two distinct obligations. A late recovery shipment does not create another denominator row. Aggregate its receipt events under the existing ID. Define any early-delivery allowance, quantity tolerance, quality condition, cancellation and authorized requirement-change policy before using real records.
Worked example: eight obligations, three passes
These are hypothetical records, not a customer result. Every obligation is for 100 units. There are no cancellations, customer changes, quality holds, overdeliveries or unit conversions. The receipt register and physical reconciliation are complete; zero means verified zero qualifying units, not missing data. All dates below are in November 2026, and every deadline is 18:00.
| ID | Original due date | Verified receipt evidence by the observation cutoff | Correct units by original deadline | OTIF pass |
|---|---|---|---|---|
| A | November 2 | R01: 100 units, November 2 at 10:00 | 100 | 1 |
| B | November 3 | R02: 100 units, November 3 at 11:00 | 100 | 1 |
| C | November 3 | R03: 60 units, November 3 at 14:00; 40 still absent | 60 | 0 |
| D | November 4 | R04: 100 units, November 5 at 09:00 | 0 | 0 |
| E | November 5 | No receipt; register reconciled at cutoff | 0 | 0 |
| F | November 5 | R05: 100 units, November 5 at 13:00 | 100 | 1 |
| G | November 6 | R06: 60 units, November 6 at 16:00; 40 still absent | 60 | 0 |
| H | November 6 | R07: 100 units of the wrong part, November 6 at 15:00 | 0 | 0 |
OTIF = passing due obligations ÷ all eligible due obligations = 3 ÷ 8 = 37.5%. Count the joint condition on each row. Do not multiply or average separate aggregate punctuality and completeness rates to estimate the observed joint count.
D is complete at the observation cutoff but missed its original deadline. C and G received some correct units on time but were short. H has a physical receipt, yet none of those wrong-part units satisfies its requirement. E has no receipt. All eight remain in this denominator.
If you keep only rows with any qualifying receipt by the observation cutoff, you retain A, B, C, D, F and G. The same three passes become 3 ÷ 6 = 50%. That is a different population: E and H disappeared. It does not show better performance against the eight obligations due.
In a spreadsheet, put the eight pass flags in E2:E9 and use =SUM(E2:E9)/ROWS(E2:E9), formatted as a percentage. This works because those eight rows are the complete eligible set. With filters, extra rows or missing evidence, build and validate that set first. With no obligations due, report “not applicable,” not 0% or 100%.
Keep the original result and the recovery status
The example has received 520 qualifying units out of 800 required by the cutoff. The remaining 280 are C's 40, E's 100, G's 40 and H's 100. That quantity balance is useful for recovery, but it is not the binary 3-of-8 score.
Give each failure an action and closure evidence. This proposed division of work uses the same records:
| Record | Next action and owner | Evidence needed to close the recovery |
|---|---|---|
| C and G | Purchasing checks each remaining 40-unit balance; planning checks the affected build | Verified receipt of each correct balance and the required release checks |
| D | Purchasing investigates the missed commitment with the supplier | Agreed cause and completed corrective action; receipt R04 already closes its quantity balance |
| E | Purchasing obtains a dated, quantity-specific recovery proposal; planning assesses the shortage | Actual receipt and required release evidence, not just a new promised date |
| H | Receiving and quality contain the wrong part; purchasing arranges the authorized replacement/return | Correct replacement receipt, required release and disposition evidence for the wrong goods |
Suppose E's supplier proposes November 9. Record the proposal and its source, assess it and obtain the required approvals. It is still a proposal until agreed. Even an approved new date leaves the original 3-of-8 result unchanged.
For a hypothetical continuation, assume the correct 100 units arrive on November 9 at 11:00 under R08 and the required release checks pass. E's recovery can then close with that evidence. Its original November 5 deadline remains a failure. Keep the historical measurement and current recovery state separately. If original data was wrong, retain a versioned correction and explain any restatement.
Connect supplier recovery to the customer decision
A supplier's new date is an input to planning. Check usable stock after other allocations, the remaining components, production capacity, testing, handling and transit before proposing a customer date. The responsible commercial owner coordinates any customer change through the agreed approval process. The supplier's OTIF score alone neither proves current readiness nor authorizes a customer commitment.
Use the sales-order change guide for the customer decision and the fulfillment guide for release, dispatch and delivery evidence. Record which customer order each shortage exposes; do not infer the cause or business impact from the aggregate percentage.
Cartograph's supplier delivery workflow supports confirming dates, checking commitments and following up on missing shipments, with communication retained on the task and approvals governed by your settings. Keep this OTIF calculation in your ERP or reporting process. The workflow evidence supports investigation; it does not establish a built-in OTIF scorecard or replace the owner's delivery decision.
Questions teams ask
What is the difference between OTIF and OTD?
OTD measures punctuality under a stated date and event rule. OTIF also requires the stated completeness condition. Define both on the same grain and baseline before comparing them; a timely partial receipt can fail OTIF.
Should a partial delivery count as a success?
Under this binary obligation method, it passes only when the whole required quantity arrives within the allowed window. Other programs award quantity-based partial credit. State which convention applies and retain any delivery schedule agreed in advance.
Can a revised delivery date improve the original OTIF result?
A supplier recovery date does not repair a missed original commitment. Track revised-plan performance separately. Handle genuine requirement changes and data corrections through a disclosed, versioned policy.
What if receipt evidence is missing?
Mark the record as a data-quality exception and investigate. Do not silently remove it, award a pass or call an unknown outcome a verified failure. Disclose incomplete evidence and withhold a definitive score if the gap could materially change it.