Orbit — Demand Forecasting Engine
Store-level demand forecasting across 1,800 locations, rebuilt on a governed data platform with a model the planners trust.
What was actually broken.
Replenishment ran on a spreadsheet model maintained by two people. It was reasonably accurate in aggregate and badly wrong at store level, so regional managers overrode it constantly — which meant the model never learned, and stock sat in the wrong buildings.
How we went at it.
We built the data platform first: governed sales, stock and promotional data with tested lineage, because a forecast on ungoverned data is just a confident guess. The model came second — gradient-boosted per-cluster forecasts with explicit uncertainty bands. Critically, we surfaced the reasoning to planners and captured every override as a labelled training signal, turning the thing planners used to do in frustration into the mechanism that improved the model.
Where it landed.
Forecast error fell by nearly a third at store level. More telling: override rate dropped by two thirds, because planners could see why the model said what it said.
The override capture was the clever bit. Our planners stopped fighting the model and started teaching it.
Vanta — Zero-Trust Platform Rebuild
Let's build the thing that moves your numbers.
Tell us what is actually blocking you. Forty-five minutes, no deck, no obligation — and a straight answer about whether we are the right team for it.