Board says 96% on time but the line still shorts—write OTIF as verifiable rules first
At the supply meeting, the big screen was all green lights, but the planner was still reading out three key parts that were out of stock. The purchaser pointed to the '96% on-time rate' saying the supplier had no problem, and the production line asked, 'Then what about the materials?' It wasn't that the data department miscalculated; 'on-time' was defined as 'having ever received goods on a certain day'.
What you'll take away
- OTIF must be assessed at the line level, and both on-time delivery and adequate qualified quantity are required, neither can be missing.
- The benchmark delivery date shall be based on the confirmed version or the procurement-accepted feedback version, and it should clearly state whether to use the original commitment or the latest version confirmed by both parties.
- Batch shipments are evaluated separately according to the plan line; completing at the end of the month does not count as punctuality throughout the whole process
- Deliveries that fail quality standards should not be counted as full quantity; early deliveries must have an allowed window, and exceeding it will be recorded as early deviation.
- OTIF needs to factor into performance and share, and be linked with order confirmation, receipt timestamps, and incoming material judgment; otherwise, the dashboard is just decoration.
The green light is bright, but the production line is still waiting for materials.
Monday supply weekly meeting, the big screen showed the supplier on-time rate at 96%. The planner opened the shortage list: three safety items ran out of stock twice this week. The purchaser said, 'The kanban is green,' the warehouse manager said, 'The truck arrived, but not the batch you need,' and quality said, 'A batch arrived but some were returned.' In the same week, from the same supplier, the three reports each told a different story.
Delivery performance distortion usually does not happen because the supplier suddenly improved, but because the assessment criteria allowed room for falsifying records. Before OTIF (On-Time In-Full) is written into verifiable rules, the ranking is only suitable for decorative purposes and not for allocating shares.
First write a page of instructions, then discuss whether to go on the big screen.
Before launching the cockpit, procurement, planning, quality, and warehouse should all co-sign a page of the 'OTIF Calibration Specification,' at least locking down these fields:
| Dimension | Recommended definition | Counterexample (Prohibited) |
|---|---|---|
| Granularity | Order line (if necessary, to the delivery schedule line) | Order header only |
| On time | The delivery timestamp falls within the promised delivery date ± allowable window | Delivery on any day of the month is considered on time |
| Commitment benchmark | Confirmed version or procurement-accepted feedback version; indicate whether to use the original confirmation or the latest mutual confirmation | Unconfirmed original date; unilateral rescheduling |
| sufficient amount | The agreed quantity of this batch is fully accepted (considering the contract tolerance) | Sending 80% can also be considered as completed. |
| Quality | Non-compliant items will not be counted in full quantity; clearly write down the concession acceptance rules | Counted upon arrival |
| in advance | Set the number of days in advance; record deviations for early occurrences beyond the window | The earlier, the better; everything is considered on time. |
| in batches | Assessments are conducted separately according to each plan; to pass the entire process, each plan must be passed. | Making up at the end of the month is considered fully on time |
Both benchmark strategies are acceptable: always evaluate performance based on the initially confirmed delivery date, or, if the customer inserts an order and both parties confirm it, evaluate based on the newly confirmed date —You can't have two sets of standards internally and one set externally.The manual can't answer everything, so don't announce the ranking yet.
Row-level computation: align three parties using a small table
Assume the order line quantity is 1000, the promised date is D, the allowed window is ±1 day, early delivery is allowed by 2 days, and quantity tolerance is 0:
| actually happened | On time | sufficient amount | OTIF |
|---|---|---|---|
| D Signed for 1000 on the day and qualified | Pass | Pass | Pass |
| D 2 Signed for 1000 qualified | Not pass | Pass | Not pass |
| D received 800, D 5 will make up another 200 | Follow the rules as planned | If not taken in full once, it is generally ineffective. | Nothing out of the ordinary |
| D received 1000, returned 50 after inspection | Pass | However (Qualified 950) | Not pass |
| D-5 Arrival Stacked at the Platform (Exceeding Early Window) | Record of early deviation | Subject to receiving policy | Usually not considered pass |
| Both plans operate at 500 each, and both meet the standards within the window. | Pass line by line | Pass line by line | Strategy passed |
The premise of 'conducting assessments according to the plan' is that the delivery plan exists after being confirmed by both parties, rather than being fabricated by the warehouse management after the goods arrive. Post-arrival fabrication is another form of padding.
Four types of water injection, more common than the formula itself
| Technique | How to identify | How to block |
|---|---|---|
| Expand the time window | The window changed from ±1 day to ±7 days, and the achievement rate surged | Window changes go through approval, historical recalculations leave traces |
| Eliminate difficult orders | The denominator becomes smaller, the numerator stays the same | Excluding required reason codes and permissions, the monthly report discloses the exclusion proportion |
| Even if it's unqualified, it will still be sent. | Return still considered completed | Sufficiently bound qualified receiving quantity |
| Use the shipping date instead of the receipt date | In transit 'On time' | Unified receipt timestamp |
Extra anti-patterns: replacing line items with the order header; diluting key components with low-value lines; having the purchasing agent change the promised date to 'protect ranking'. Useful accompanying signals: line-level OTIF, key component OTIF, early deviation rate, number of confirmed delivery date changes, proportion of supplier responsibility in material shortage work orders.
OTIF is a byproduct of the three links, not a reporting side job.
| Step | Facts that need to be kept | The significance of OTIF |
|---|---|---|
| Order Confirmation / Reply | Entering the industry after confirming the version and the promised delivery date | Reliable base date |
| Shipping plan | Planned Number of Trips and Expected Arrival | Can be calculated separately in batches |
| Appointment and Receipt | Check-in / Delivery Timestamp and Actual Received Quantity | The truth about punctual time |
| Incoming Material Inspection | Qualified / Unqualified / Accepted Quantity for Concession Rework | A sufficient amount of mass truth |
| Performance and Share | Line-level OTIF improvement and share cap | Only the signboard has teeth |
Randomly select a row of overdue materials. If you cannot drill down to 'Promise Date Source, Receipt Time, Actual Qualified Quantity, Unqualified Quantity, Batch Plan Line,' then do not publish the ranking. When the order confirmation rate is too low, the reference date is unreliable—first fix the confirmation and feedback before discussing OTIF. When receipt timestamps are mostly recorded afterward, on-time performance loses its meaning—first enforce real-time posting discipline.
In the supply plan, delivery coordination and performance are usually counted by line for on-time delivery, and are connected with order confirmation, delivery response, and portal shipment; only after warehouse receipt and returned defective materials are written back does OTIF align with the production line's perception. When JEDP-SRM observes line-level metrics and links share/rectification actions, priority is given to firmly drilling down on capabilities, rather than first making a more visually appealing large screen.
Announce the rankings first, pass through three thresholds
- Confirmation rate thresholdSuppliers whose strategic material order confirmation rate does not meet the standard will only have internal shadow reports issued and will not be included in the public ranking.
- Timestamp thresholdWhen the sampling inspection signing time deviates too much from the gatekeeper/appointment check-in time, the data ranking for that warehouse will be suspended.
- Quality Feedback ThresholdWhen test results cannot be written back to the received quantity, full quantity is always regarded as unavailable to avoid 'full score upon delivery'.
Share linkage is recommended to start gradually: if it falls below the threshold for two consecutive periods → corrective tasks; if it still lags at the row level for three periods and the reason is with the supplier → downgrade new business share. During the calibration period, implementing 'bottom elimination' immediately will only lead to more serious inflating of results. During the shadow report stage, it is only internal; the supplier does not yet see the ranking, so first run through explanations for the random sampling of fifty rows.
The implementation sequence can be condensed as: define the freeze in writing → align confirmation rates → ensure shadow report sampling is explainable → publish the ranking and link it to rectification/shares. If the metrics are undisputed for two consecutive months, set further improvement goals.
Don't forcefully apply traditional PO to OTIF in these scenarios for now
- Consignment / VMI and consumption-triggered replenishment: switch to replenishment response timeliness and inventory health.
- Responsibility for material shortages is unclear (design changes, urgent customer orders, and warehouse capacity rejections are mixed together): first, determine the cause code;
- Ranking the group without unifying multi-factory standards: this will penalize factories with stricter definitions;
- The internal plan itself swings on a weekly basis but requires the supplier to confirm instantly: the portal will become a tool for mutual harm.
There is no universal passing standard. First, ensure the specifications are stable for three months, then set targets according to material loss and product category; before comparing across factories, the definitions in the manual must be standardized.
The most common mistake during the definition alignment period is modifying the window while simultaneously publishing the ranking. Suppliers may think the rules are negotiable, reducing their motivation to make corrections. Any changes to the specifications must announce the effective date, and it should be clearly stated whether historical data is recalculated; manual adjustments to individual rows can remain as an option, but the reason, approver, and validity period must be provided—if adjustments become the default, the row-level dashboard will revert to being decorative.
Do not crowd critical parts and long-tail parts into the same 'average on-time rate' to show off. Safety parts and long-lead-time parts should report critical part OTIF separately; low-value auxiliary materials can be weighted by amount, but do not use green for auxiliary materials to cover up red for safety parts. In meetings, first read the shortage list of critical parts, then look at the total score—if the order is reversed, you'll go back to '96% still short of materials'.
What to do next
This week, write a one-page OTIF specification manual (line level, baseline date, window, lead time, quality, batch), to be countersigned by the four departments. Before countersigning, randomly select fifty lines of historical deliveries and test calculate them according to the manual to see how many lines are 'kanban green, production line red.' For fields that cannot be clearly explained, first revise the manual, then decide whether to display them on the big screen or include them in the share allocation meeting.
- SRM Supply Solutions: Collaboration and Performance
- How to reconcile and track in-transit goods based on delivery facts
- Purchase Order Collaboration: Confirmation, Changes, and Feedback
- Delivery Appointment and Platform Occupancy
- Supplier Performance and Share Exit
Want to apply this in your plant?Resources has matching self-assessments and templates—or book a live demo.
FAQ
Quick answers on this topic—still stuck?Contact
Is early delivery considered on time or a violation?
It depends on whether early arrivals are allowed. It is recommended to set the number of days allowed for early arrival; arrivals within the window are counted as on time, while those that exceed it are recorded as 'too early' deviation rather than OTIF pass. When warehouse capacity is tight, being too early is as harmful as being too late.
How is the delivery time change caused by a customer rush order calculated?
Both parties have confirmed the assessment according to the new delivery schedule; if the main factory unilaterally changes the schedule without the supplier's confirmation, the supplier should not be penalized. The responsibility lies in the change confirmation process, not in the OTIF formula itself.
How to summarize multiple materials?
At the same time, look at project OTIF in terms of either quantity or amount, or weighted OTIF for key items, to avoid being diluted by a large amount of low-value materials. Key items can have a separate board.
What counts as passing?
There is no universal threshold. First ensure that the definition is stable for three months, and then set improvement goals according to the loss of missing materials and categories; it must be uniformly defined before cross-factory comparison.
Should it be considered on-time based on the shipping date or the receiving date?
It is recommended to take the warehouse receipt timestamp as the standard. The shipping date can be declared unilaterally by the supplier, while the receiving date aligns with the availability of the production line.