How to raise fill rate without buying more stock

First, make sure you are measuring the right thing

Before any improvement work, settle three questions, because each of them can shift the reported number by several points without anything changing on the floor:

• Which fill rate? Line, unit or order. They are different numbers. Pick one as the headline and report the other two beside it. Calculate all three.
• Against which quantity? The originally requested quantity, not a revised one. If customer service reduces a line to match available stock, that line is not filled.
• VOR separately from stock orders. Emergency and machine-down lines behave completely differently from routine replenishment. Blending them hides both problems.

Getting these wrong wastes the first month of any improvement programme arguing about whether the baseline was real.

Lever 1 — Fix inventory accuracy

Cost: low. Speed: 4–8 weeks. Typical gain: 2–5 points.

A stockout caused by stock that the system thinks exists is not a planning failure, it is an accuracy failure, and no amount of buying fixes it. If location-level accuracy is below 97%, this is the first job and nothing else will hold until it is done.

Blind counts only. Root-cause every variance above a threshold rather than adjusting it away — the adjustment closes the variance and destroys the information. The causes are nearly always a short list: picking from the wrong location, put-away to an unrecorded location, damage not booked, or a receipt never posted. Each has a specific process fix. Size the counting programme.

Lever 2 — Correct min-max on the failing items

Cost: low. Speed: 2–4 weeks. Typical gain: 3–6 points.

This is the highest-return work available, and it is usually a few days of effort. Take the top group of failing SKUs from your miss list and recalculate their levels properly — using the full safety stock formula that accounts for lead-time variability, not the simple one. Run the numbers.

Two things come out of this consistently. First, many of the failing items have levels set years ago against lead times that no longer exist. Second, the buffer on several of them is driven by lead-time variability rather than demand variability — which means the fix is a supplier conversation, not a purchase order.

Lever 3 — Over-stock the cheap, predictable tail

Cost: low. Speed: one replenishment cycle. Typical gain: 2–4 points on order fill.

The most counter-intuitive lever, and the most reliably effective on order fill. CX items — low value, steady demand — cost almost nothing to hold. A part costing ₹40 with a holding cost of 22% costs under ₹9 a year to carry. Holding three months of cover instead of three weeks is a rounding error on the inventory value.

But because nobody pays attention to cheap items, they stock out regularly, and each stockout spoils a whole customer order. This is why order fill is so often far below line fill. Classify the catalogue, find the CX group, and be generous with them. Classify it.

Lever 4 — Attack lead time, not just stock

Cost: medium. Speed: 2–6 months. Typical gain: 2–5 points, and it lowers inventory.

This is the only lever that improves service and reduces stock at the same time, which is why it is worth the effort even though it is slower.

Safety stock scales with the square root of lead time and directly with lead-time variability. Halving the variability on a long-lead item can release a quarter of its buffer while improving availability. The work is unglamorous: measure actual lead times from your own GRN data rather than trusting quoted ones, find the items where actual and quoted diverge most, and take that evidence to the supplier.

Do not overlook the internal portion. Dock-to-stock is part of your real lead time. A 48-hour dock-to-stock on a 21-day supply lead time is a 10% extension of it, entirely within your own control. Measure it.

Lever 5 — Then, and only then, buy more stock

Cost: high, and recurring. Speed: one lead time. Typical gain: diminishing.

After the first four levers, whatever gap remains is genuine and needs inventory. By that point you will be buying cover for a specific, evidenced list of items rather than spreading money across the catalogue — which is the difference between a ₹10 lakh intervention and a ₹1 crore one for the same result.

Set the service level by class rather than applying one target to everything. 98–99% on AX and critical VOR items, 95% on B, 90% on the tail. A uniform 98% across a long-tailed spare-parts catalogue is enormously expensive and delivers almost nothing on the items where it costs the most.

A 90-day sequence

What to watch while doing this

Fill rate can always be bought. Report inventory turns and ageing stock on the same page, every month, throughout the programme. A fill-rate improvement accompanied by a collapse in turns is not an improvement; it is a decision to spend working capital, and it should be taken deliberately and visibly rather than by accident.

The honest measure of success is fill rate up and turns flat or better. That combination is achievable — but only if the first four levers are pulled before the fifth.