“Hold forty-five days of cover” is the kind of instruction that feels like control and behaves like the opposite. It is clean, it is auditable, and it is wrong almost everywhere at once - too much stock on the items that sell predictably, too little on the ones that do not. A flat target treats inventory as a policy to be set. It is better understood as an output: the consequence of a service-level promise, demand variability and lead time, computed per item and refreshed as those inputs move.
1. A flat days-of-cover target is a hidden service-level decision
Days of cover looks like a stock decision, but it silently sets a service level - a different one for every item, none of them chosen. For a steady, predictable SKU, forty-five days might deliver 99% availability and a pile of dead capital. For an erratic one with the same forty-five days, it might deliver 88% and regular stock-outs. You did not decide either of those outcomes; the flat number decided them for you, badly, behind your back. The first move is to invert the logic: decide the service level you actually want - which can and should differ by segment - and let the cover fall out of it.
2. Variability and lead time belong in the number
The reason one target cannot fit two items is that safety stock is driven by things a days-of-cover figure ignores entirely: how variable demand is, how long and how reliable the replenishment lead time is, and how confident you want to be against both. Two SKUs with identical average sales can need very different buffers if one is smooth and locally sourced and the other is lumpy and shipped from the other side of the world. A real target carries those drivers inside it, which is precisely why it has to be calculated rather than declared - no committee can hold the variance of ten thousand items in its head, but a model holds it trivially.
3. The long tail is where policy targets do the most damage
Flat targets fail most expensively in the long tail, and the long tail is usually most of the catalogue. Slow, intermittent items break the assumptions a single cover number quietly relies on; applying the fast-mover policy to them is how you end up simultaneously writing off obsolete stock and missing sales on the same shelf. The tail needs its own logic - intermittent-demand methods, deliberate stock-or-not decisions, sometimes a make-to-order stance - not a smaller slice of the same blanket rule. Segment first, then let each segment compute its own target; the gains concentrate here.
4. Make the target a living output
Even a well-calculated target decays. Demand variability shifts with the season, suppliers re-rate their lead times, service ambitions change with the commercial strategy. If the target is a number typed into a policy document once a year, it is wrong for eleven months of every twelve. Treated as an output, it is recomputed every planning cycle from current inputs and flows straight into replenishment - so the buffer breathes with the business instead of lagging a year behind it. This is also where a connected planning model earns its place: the service level, the statistical drivers and the resulting target live in one structure - Anaplan is well suited to it - so a change in assumption propagates to the buffer and the working-capital number in the same cycle.
Where to start
- Replace the flat days-of-cover rule with explicit service-level targets by segment - that single reframing exposes most of the mispriced risk.
- Compute safety stock from demand variability and lead time, per item, rather than declaring cover.
- Give the long tail its own logic; do not shrink the fast-mover policy and hope.
- Recompute targets every cycle and feed them into replenishment, so the number stays alive.
The prize is the one every supply chain is told it cannot have: less inventory and better availability at the same time, because the stock moves off the items that never needed it and onto the ones that did. An engagement to rebuild targets this way is usually four to eight weeks depending on catalogue complexity - and the working capital it releases tends to pay for itself well inside the first year.
