Fill Rate Math: Why 95% Costs You More Than You Think

The 95% illusion

A 95% fill rate sits in most retail scorecards as "green." Industry benchmarks say 95-97% is healthy. Procurement teams celebrate it. Operations teams report it monthly. And it is one of the most expensive metrics in retail to misread.

The number itself is fine. The problem is what it averages over. Fill rate is the ratio of units shipped (or units available on shelf) to units ordered or demanded. At 95%, the 5% gap looks small. It isn't. Five percent of demand, multiplied by the basket effect, multiplied by walked customers, multiplied by category substitution drag, lands on the P&L as 1.5-3% of revenue gone.

For a $400M chain, that's $6M-$12M per year. The scorecard says 95%. The P&L says something different.

The basket effect

A missed unit isn't a missed unit. A missed unit is, on average, a missed basket. Customers in multi-category retail don't visit for one SKU. They visit with a list. The grocery shopper has 14 items. The home improvement shopper has 6. The specialty apparel shopper has 3-4 intent items plus discovery.

When the first item is out, the math changes immediately. POS data across mid-market chains shows that 22-38% of customers who hit a stockout on a planned item leave without completing the rest of the basket. They go to a competitor for the missing item and complete the full list there. The retailer didn't lose one unit. They lost the whole trip.

That's the multiplier most fill rate calculations ignore. A 5% unit-level stockout rate creates roughly a 1.5-2.5% trip-level abandonment rate, weighted by the position of the missing item in the basket and the basket composition.

For a chain with a $42 average basket and 18M annual transactions, even a 1.8% trip-level abandonment means 324,000 lost trips. At $42 each, that's $13.6M in lost revenue from what the scorecard categorizes as a healthy fill rate.

Category substitution drag

Some stockouts don't lose the trip. The customer substitutes. They pick the alternative brand, the alternative size, or the alternative SKU. Procurement teams sometimes treat this as a save. It usually isn't.

Substitution drag has three sources of cost. The first is margin mix. The substituted SKU is rarely margin-equivalent to the intended one. In categories where private-label sits next to premium national brands, an out-of-stock on the premium SKU pushes customers to private-label, which is typically 8-15 margin points lower for the retailer but usually at a lower price point. Net: lower absolute margin dollars per unit.

The second is satisfaction decay. Customers who substitute report lower satisfaction scores for the trip even when they complete the basket. NPS studies on substitution events show 12-18 point drops at the trip level. Repeat trip frequency from those customers drops 6-9% over the next 90 days.

The third is the silent attrition cost. The customer who substitutes once is more likely to substitute permanently, especially when they substitute to a competitor's brand carried at your store. You spent shelf space and replenishment cycles building demand for the SKU you couldn't fulfill, and converted that demand into preference for the SKU you stock alongside it. The original brand's velocity decays, and the next quarterly review questions why you carry it.

Quantifying the substitution cost

For a typical grocery or convenience retailer, 30-50% of stockout events result in substitution rather than walked baskets. Of those substitutions, roughly 60% are to lower-margin alternatives. The average margin gap on substituted units is 4-7 points.

Run the math on a $400M chain with 5% fill rate gap: roughly $20M in missed demand. Forty percent substitutes, two-thirds of those substitute to lower margin. That's $5.3M flowing through at a 5-point lower margin. The annualized margin drag is $266K from this single failure mode, before counting the satisfaction decay and attrition.

Most fill rate scorecards don't capture this because they stop at the unit count. The unit was "fulfilled" via substitution. The margin damage doesn't show up until the quarterly margin variance review, by which point it's been attributed to "category mix shift" rather than to fill rate.

How to calculate fill rate, and the three versions people confuse

Half the arguments about fill rate are actually arguments about which fill rate. Three definitions circulate, they produce different numbers off the same data, and teams routinely compare one to another without noticing.

Unit fill rate. Units shipped divided by units ordered, times 100. Order 1,000 units, receive 940, and unit fill rate is 94%. This is the most forgiving version because a large shipment of one SKU can mask complete failure on another.

Line fill rate. Order lines filled completely divided by total order lines. Twenty lines on a purchase order, seventeen arrive complete, and line fill is 85% even if unit fill reads 94%. This version is harsher and closer to what the shelf experiences, because a partially filled line still leaves a gap.

Case fill rate. Cases shipped divided by cases ordered. Used mostly in supplier scorecards and DC-to-store replenishment. Useful for vendor management, close to meaningless for customer experience.

The retail version that matters to the customer is none of these three. It is on-shelf availability: the percentage of expected SKUs physically present and buyable at the moment a customer reaches for them. A DC can hit 98% case fill while a store sits at 89% on-shelf availability, because the units arrived and never made it to the shelf.

If your supplier scorecard says 97% and your customers keep finding holes, you are not looking at a contradiction. You are looking at two different metrics measuring two different handoffs.

Fill rate benchmarks by vertical

Benchmarks are only useful when they are specific to the format and the SKU tier. A single chain number compared against a single industry number tells you nothing worth acting on.

Notice the second column. In every format the standard for top-velocity SKUs runs several points above the blended number. That is the whole argument in one line. The blended benchmark is satisfied by slow movers you always have in stock, and the top movers carry the trips.

Pharmacy runs the tightest standard for an obvious reason: a customer who cannot fill a prescription does not substitute, they transfer, and the transfer is usually permanent. Fashion tolerates a lower blended number because breadth is part of the proposition, but core sizes behave like grocery staples.

Diagnosing where your fill rate is actually breaking

A fill rate gap has four possible homes, and the fix is different in each. Working through them in order stops teams from throwing inventory at a problem that inventory does not solve.

Supply. The units never arrived. Check supplier on-time-in-full before anything else, because no store-level intervention fixes a vendor that ships short. The OTIF scorecard is the diagnostic here.

Allocation. The units arrived at the chain but went to the wrong stores. This shows up as high fill rate variance across stores with similar demand profiles. Central allocation running on stale demand assumptions is the usual cause.

Replenishment timing. The units are in the building and the shelf is empty anyway. Backroom stock with an empty facing is the single most common failure in multi-store retail, and it is invisible to any system that reads inventory position rather than shelf state.

Data. The system believes stock exists that does not. Phantom inventory stops the replenishment engine from reordering, so the gap persists indefinitely. This one compounds: every day the count stays wrong is another day nothing gets ordered. See phantom inventory detection.

Run them in that order. Most chains that start by raising safety stock are treating an allocation or data problem with working capital, which is why the fill rate barely moves and the inventory line jumps.

Lane-level fill rate is the only honest version

Chain-level fill rate of 95% is an average across stores that ranged from 88% to 99%. The chain reports the middle and operates around the middle. But the customers who walked are not distributed evenly. They concentrate in the stores at the bottom of the distribution.

One regional grocer with 90 stores reported 95.4% chain fill rate. Decomposed by store, 12 stores were below 92%. Those 12 stores accounted for 64% of the chain's lost-basket events. They were also the stores with the most price-sensitive customer demographics, the most competitive radius density, and the lowest tolerance for substitution. The chain was losing customers in exactly the markets where customer acquisition cost was highest.

The decomposition doesn't stop at store. It needs to go to category and daypart. A 95% chain fill rate often means 99% on slow movers (which always have plenty of inventory) and 86% on top-velocity SKUs during peak hours. The slow movers don't matter to customers. The top movers are the trip drivers. The fill rate gap concentrates exactly where it hurts most.

Honest fill rate measurement looks like this: top 200 SKUs by velocity, by store, by daypart (open-12pm, 12pm-5pm, 5pm-close), by day of week. That's the cube where the customer experience lives. The chain-level number is a marketing artifact, not an operating signal.

Signal-based fill rate recovery

Fill rate recovery doesn't require more inventory across the board. It requires more inventory in the specific store-SKU-daypart cells where customer impact concentrates. Most chains overstock 60% of their inventory and understock 8-12%. The 8-12% is where the fill rate damage lives.

The mechanics: continuous signal monitoring on POS velocity, replenishment timing, and on-hand variance flags the SKU-store combinations that are trending toward stockout 24-48 hours before the gap appears. Replenishment cycles get adjusted at the store level rather than centrally. Safety stock floors get re-baselined on top movers in high-impact stores while being lowered on slow movers elsewhere.

The retailers running this pattern improve fill rate on top-200 SKUs from 92-94% to 98-99% without increasing total inventory dollars. They reduce slow-mover inventory at the same time. Working capital is flat or better. Fill rate at the customer-relevant cube improves dramatically.

The financial outcome is consistent: 1.2-2.4% comp lift from recovered demand, 60-90bps margin improvement from reduced substitution drag, and a 15-25% reduction in safety stock on the slow-mover tail. For a $400M chain, that's $5-10M in recovered revenue plus $2-4M in working capital release.

Key takeaways

• Chain-level fill rate of 95% routinely hides $6M-$12M per year in lost revenue and substitution drag for a $400M retailer.
• 22-38% of customers who hit a stockout abandon the entire basket, not just the missed item. Unit-level fill rate misses this multiplier completely.
• Substitution events cost margin even when they "save" the unit count: lower-margin alternatives, 12-18 point NPS drops, and 6-9% repeat trip decay over 90 days.
• The chain fill rate average hides 12-15% of stores running below 92%: exactly the stores where customer acquisition cost is highest.
• Unit fill, line fill, and case fill produce different numbers from the same data. None of them is on-shelf availability, which is the only version the customer experiences.
• Benchmarks only mean something by format and SKU tier. Top-velocity standards run several points above the blended number in every vertical, and those are the SKUs that carry the trips.
• Diagnose the gap in order: supply, allocation, replenishment timing, then data. Chains that start by raising safety stock are usually treating an allocation or phantom-inventory problem with working capital.
• Honest fill rate is measured at SKU-store-daypart granularity on top-velocity items, not as a chain-level monthly average.
• Signal-based replenishment recovers fill rate on top movers without increasing total inventory dollars, and usually reduces slow-mover overstock at the same time.
• The path from 94% to 99% fill rate on the top-200 velocity cube is typically worth 1.2-2.4% comp lift plus 60-90bps of margin recovery.