In a configure-to-order business a perfect order touches no human between the dealer's click and the plant's schedule. Fallout is every order that doesn't make it. Each one costs someone's time, and many cost a delivery date.
Where it comes from
| Cause | Looks like | Real fix |
|---|---|---|
| Missing constraint | An unbuildable combination was allowed | Add the rule and a test |
| Stale orderability | A retired part is still selectable | Lifecycle state and effectivity |
| Pricing gap | An option with no price | Publish rules and prices together |
| Master data | Wrong unit, code or description | Data governance |
| Timing | A change landed mid-order | Better cut-in dates |
The habit that matters
Fixing the order is the easy half. The useful half is asking what let it through and fixing that, then saving the failed configuration as a test so it can't return. I was the single-point owner of order fallout for North American B2B dealers and OEMs at MillerKnoll, and working this way is part of how we reduced quoting errors by 30%. The longer version is in Order fallout is a product problem.
Questions people ask
Is some fallout normal?
Yes. Truly special requests will always need a person. The goal is that what falls out is genuinely special, not a repeat of last week's error.
How should it be measured?
As a rate (orders needing intervention over all configured orders), by cause, and by repeat cause. Repeat fallout is the most telling number.
Updated November 2026. Think I've got something wrong? Tell me.