Quota
An allocation service that split incoming orders across factories under capacity and lead-time constraints, and — the part that mattered — wrote down why each order landed where it did.
- Year
- 2023
- Role
- Build
- Built with
- Go, Postgres
The problem
The original allocator was correct and completely opaque. When an order went to a slower factory, nobody could reconstruct the reason, so operators stopped trusting it and started overriding by hand.
Approach
Every allocation emits the binding constraint alongside the decision: which capacity ceiling or lead-time floor forced it. Override rates dropped once operators could see that the scheduler already knew what they were about to tell it.
Why archived
Superseded when order volume outgrew the assumption that allocation could run synchronously. The explanation format survived into its replacement, which I think was the actual contribution.