CPG Scheduler
All posts

OTIF Starts on the Line: Scheduling to the Must-Arrive-By Date

schedulingstockoutscpg operationscapacitysequencing

On-Time and In-Full — Why OTIF Is Won on the Production Line

Here is a fact that reorders how you should think about a due date: a major retailer can fine you for arriving early just as readily as for arriving late. Walmart's "On-Time, In-Full" program charged suppliers a penalty equal to 3% of the value of goods that were late or missing — and it penalized early shipments too, because arriving ahead of schedule creates overstocks and clogs the distribution center (IndustryWeek).

That single detail kills a comfortable assumption. The ship date is not the target. The must-arrive-by date (MABD) is the target — a narrow window at the retailer's dock that you have to hit from the inside, which means the real deadline lives back on your production line. OTIF looks like a logistics metric. It is actually a scheduling metric wearing a logistics costume.

This piece covers what OTIF measures, how it's calculated, what a miss costs, and the concrete scheduling levers that protect it.

What OTIF actually measures (and why it's harsher than it looks)

OTIF — on-time in-full, also called DIFOT (delivery in-full, on-time) — measures delivery performance from the customer's point of view: the right product, in the ordered quantity, at the agreed place, at the time expected, usually within a customer-defined tolerance window, expressed as a percentage (Wikipedia, DIFOT).

That customer's-eye framing is the whole point. OTIF is considered superior to "shipped-on-time" (SOT) or on-time performance (OTP) precisely because those metrics are judged from the shipper's perspective — they reward you for pushing product out your own door. OTIF only counts what the customer actually received, on the terms the customer set.

And it is unforgiving in a way that trips up operators who think in averages. OTIF is binary. Delivering 98% of the ordered quantity on time is still a miss, because the in-full component is all-or-nothing. Measuring OTIF by the percentage of order quantity delivered actually violates the principle — a partial-but-punctual line is simply not OTIF (Wikipedia, DIFOT). You either met both halves, or you failed the line.

Expect a naming zoo, too. The same idea shows up as OTIF, DIFOT, On-Time Fill Rate (OTFR) at Target, and Original Requested Arrival Date (ORAD) at Kroger. An extended version, DIFOTAI, adds "accurately invoiced." Different label, same unforgiving scorecard.

How OTIF is calculated — pick your denominator carefully

The arithmetic is simple; the denominator is where you get fooled. Organizations compute OTIF over deliveries, orders, or order lines, and split order lines are each counted separately (Wikipedia, DIFOT). The base you choose changes the number dramatically.

Work a quick example. Suppose a purchase order has three order lines, and you ship two of them complete and on time but short-ship the third:

  • Measured by quantity: you might report something flattering like 90%+.
  • Measured by order lines: two of three lines pass → 67%.
  • Measured by whole order (all-or-nothing): the order is not complete → 0%.

Same shipment, three very different scores. Leading UK-retail practice, per the Godsell and van Hoek research cited in the DIFOT literature, demands OTIF in excess of 97% measured at the SKU level — the strict, line-level base, not the forgiving one.

Then there's the request-date-versus-promise-date trap. If you score yourself against your own promise date, you can look terrific — and then watch the number collapse when the retailer re-measures against their requested date (Wikipedia, DIFOT). The retailer grades to the request date. If your internal scorecard quietly uses the promise date, you're flying blind to your real compliance. This is the downstream bookend to order promising: Available-to-Promise quotes a date; OTIF grades whether you hit it.

The stakes — what a scorecard miss costs

Walmart's program makes the stakes concrete. For full-truckload suppliers of fast-turning items, the initial rule was to deliver 100% in full, on the must-arrive-by date, 75% of the time, with the target climbing to 95% — up from a prior 90% measured against a more lenient four-day window (IndustryWeek). The window tightened and the bar rose.

Miss it and you pay the 3% chargeback on late or missing goods — and, again, early shipments get penalized because they create overstock and DC congestion. The program was built to add roughly $1 billion in revenue through better on-shelf availability, which tells you how seriously the retailer treats availability.

Now the reality check. When the program launched, OTIF for Walmart's top 75 suppliers — household names including P&G and Unilever — had been as low as 10%, and none had reached the 95% long-term target (IndustryWeek). These are the most sophisticated supply chains in consumer goods, and they were starting from single digits. If that's the baseline, the problem is structural, not a matter of a few bad weeks.

The leverage is enormous because of scale and concentration. Walmart's network exceeded 150 U.S. distribution centers, and a single retailer can represent a large chunk of a supplier's sales — on the order of a quarter of total revenue for some vendors. A compliance program at that customer isn't a line item; it's a board-level number.

OTIF is a scheduling problem before it's a logistics problem

Walmart's own diagnosis of the root cause was blunt: "Variability is the No. 1 killer of the supply chain" (IndustryWeek). That is a scheduling statement. Variability in your run rates, your changeovers, and your yields is what blows the arrival window.

Here's the mental model. Start at the MABD and subtract backward:

MABD − transit − dock/stage − pick/pack − production finish = your line's true due date.

That production-finish timestamp is the deadline your schedule actually owns. Everything downstream of it is relatively deterministic; everything upstream of it is where your variability lives. If your finish time is honest, the rest of the chain has room to work. If it's optimistic, you've already lost the window before the truck is loaded.

And don't forget the in-full half. Completeness is a production constraint, not a warehouse one. It ties directly to batch sizing, yield, and short-run risk — the same tensions covered in the newsvendor one-shot run and economic production quantity pieces. If a run comes up short on yield, you don't just have an inventory problem — you have an OTIF-zero on that line.

The benchmark literature agrees on the linkage. ISM frames delivery reliability as something that "protects production schedules and service levels," explicitly tying OTIF to the scheduling function, and notes that measuring the spread of lead time — not just the average — is what prompts corrective actions like capacity adjustments. Average on-time means nothing if the variance means one order in five lands outside the window.

Five scheduling levers to protect OTIF

1. Schedule to the MABD-derived line due date — and buffer for variability, not with a flat pad. Back-plan from the arrival window to the production-finish deadline, then size the lead-time buffer to the spread of your process, not a gut-feel number of days. The safety-stock formulas that account for lead-time variability carry the same logic into the time dimension: variability, not the mean, is what you're protecting against.

2. Protect the in-full component — don't release a short run. Because in-full is all-or-nothing, sequence changeovers so the full order-line quantity completes before the cutoff. A line that finishes 90% of the quantity on time scores zero. Faster, more reliable changeovers buy you the schedule room to finish complete; see SMED changeover reduction and sequence-dependent changeover scheduling.

3. Freeze orders near the MABD. Last-minute reshuffles are how in-window orders become late ones. A time-fence / frozen zone near the arrival date keeps someone's urgent expedite from knocking a committed order out of its window.

4. Use real, OEE-based run rates so the finish time is honest. If your back-plan assumes nameplate speed, your production-finish timestamp is fiction. Plan to the schedulable run rate derived from OEE, not the theoretical maximum, so the MABD math holds up on the floor.

5. Instrument every miss with a reason code. Capture why each non-OTIF line failed — supplier-fault versus internal, short-yield versus late-start versus transit — so the scorecard drives corrective action instead of blame. ISM's framing is that the scorecard should show both punctuality and completeness so you can act on the pattern.

OTIF vs. the service metrics you already track

OTIF is not a replacement for the inventory-sizing metrics you already run — it's the customer-facing complement to them.

Metric What it grades Base Logic
OTIF / DIFOT Did the customer receive it, complete, in-window Order line (binary) Compliance
Fill rate Fraction of demand filled from stock Units Inventory sizing
Cycle service level Probability of no stockout per cycle Cycles Safety-stock sizing

Fill rate and cycle service level size your buffers; OTIF grades whether the retailer actually got what they ordered on the day they wanted it. The distinction between the first two is unpacked in cycle service level vs. fill rate. OTIF sits on top: you can run a healthy fill rate internally and still fail OTIF if your timing is loose.

A simple OTIF scorecard to run on your line

You don't need the retailer's portal to start. Build your own:

  1. Define the formula and denominator you'll own — commit to order-line, SKU-level measurement so you're grading yourself the way the retailer does.
  2. Measure against the requested date, never your own promise date, so your number reflects reality.
  3. Set the target. ISM cites a general on-time benchmark of at least 95% (OTIF where applicable) with order accuracy of at least 99%; strict retail practice pushes OTIF above 97% at SKU level.
  4. Run it monthly, and trend the spread of your arrival timing, not just the average.
  5. Tie every miss to a reason code and a corrective action — a capacity adjustment, a changeover resequence, a tighter time fence.

Key takeaways

  • The MABD, not the ship date, is the real due date — and early can be penalized as hard as late.
  • In-full is all-or-nothing; a short-shipped line scores zero no matter how punctual it is.
  • Measure at SKU/order-line level against the requested date, or you'll flatter yourself into a chargeback.
  • OTIF is driven by variability, which makes it a scheduling problem: back-plan from the arrival window to an honest production-finish deadline and buffer the spread.
  • Reason-code every miss so the scorecard drives changes on the line.

CPG Scheduler back-plans from the must-arrive-by date to the production-finish deadline and flags at-risk order lines — short yields, late starts, tight changeover windows — before they become chargebacks. See how an agent-ready scheduler works.

Sources

More from the journal