An auto-parts distributor carries the right part for thousands of vehicles, most of which sell a handful of times a year, while promising it on the shelf the next day. That promise is expensive to keep badly.
Sources: Advance Auto Parts FY2024 Form 10-K; Genuine Parts Company FY2024 Form 10-K (SEC).
In fiscal 2024 two large parts distributors booked real inventory pain in their own filings: one took a $431.5 million inventory write-down tied to store closures and liquidation, another a $62 million charge driven by obsolescence. Those are not freak events. They are what happens when a business model requires holding an enormous fitment catalogue, most of it slow, against a promise of next-day availability. Get the sizing wrong across that breadth and the charges show up at year-end.
A parts catalogue runs to tens or hundreds of thousands of SKUs, driven by vehicle fitment. The large majority sell a few times a year. Run a moving average over a part that sold twice and it rounds the forecast toward zero, so you either hold it on instinct or drop it and lose the sale that mattered. Intermittent demand is the rule here, not the exception, and classic forecasting has almost nothing useful to say about it.
Customers expect the exact part, in stock, tomorrow. That service promise forces buffers on slow movers that would otherwise be dropped, which is where the cash goes. The only honest way through is a per-part buffer sized on that part's real demand and lead-time variability, and on how critical availability truly is, high where a miss loses the customer, lean where a short wait is fine. A blanket service target across the catalogue just funds coverage on parts nobody pulls.
Parts get superseded and fitment changes, so the sales history behind a forecast keeps getting cut off and restarted. Plan naively on raw history and you misjudge demand right when a part is replaced. The plan has to handle supersessions and interchanges deliberately, or the forecast quietly drifts wrong across a big slice of the catalogue.
Distributors who keep the catalogue from eating their cash forecast intermittent demand part by part with a method built for it, size safety stock and reorder points on real variability and the service level each part deserves, and handle supersessions so history breaks do not corrupt the plan. They plan across branches and DCs so a part dead in one location can serve another before anyone reorders. It is the deterministic, auditable approach DemandIntel brings to automotive and mobility parts distributors, every reorder traceable, on your own data, on top of the ERP you already run.
Fitment means tens or hundreds of thousands of SKUs, most slow-moving, held against a next-day availability promise. Across that breadth it ties up large amounts of cash, as the $431.5 million and $62 million fiscal 2024 write-downs show.
With a method built for intermittent demand rather than a moving average, which rounds sporadic sales toward zero. Each slow part is then sized on its real pattern and lead time instead of guessed or over-held.
Set the buffer per part on its real demand variability and how critical availability is: high where a miss loses the customer, lean where a short wait is acceptable, rather than a blanket service target across the whole catalogue.
Supersessions and fitment changes break the sales history a forecast relies on. Handling interchanges deliberately keeps the plan accurate when a part is replaced, instead of drifting wrong across the catalogue.
A fitment catalogue is a promise to have the right part tomorrow. Keep it with a blanket buffer and you fund a warehouse of parts nobody will pull.
Picture your next S&OP meeting: one number on screen that nobody argues with. See it free, on your own data, in 45 minutes; and if it does not hold up, we part as friends.