Operational forecasting

Forecasting

Operational forecasting

Forecasts built from your own data, tested against what you use now and put into the planners' workflow.

“One person rebuilds the forecast in a spreadsheet every month. Planners change it anyway, and nobody writes down why.”

Diagram. Existing systems: Maintenance work orders, Parts usage history, Equipment fleet, Supplier lead times, Stock on hand. They feed Parts demand forecast, refined through Clean history, Forecast, Stock levels. It serves the Maintenance planners (parts at risk), the Supply chain team (forecast against actual), the Leadership team (stock and service).

Illustrative example.

  1. The problem

    “One person rebuilds the forecast in a spreadsheet every month. Planners change it anyway, and nobody writes down why.”

  2. The solution

    1. 01DecisionWhat gets ordered or held because of the forecast, and what a wrong forecast costs.
    2. 02HistorySales, orders, usage and stock, matched item by item and site by site.
    3. 03Cost of errorModels are judged by what their errors cost. Running out of a part can cost far more than a spare.
    4. 04Planner workflowPlanners see the forecast where they work, and can change it and say why.
  3. Previous work

    Supply chain forecasting platform

    For a major industrial client we designed and built a multi-tenant forecasting platform, with the models inside the teams' own workflows. It runs in production.

    Full story

    Multi-tenant means one system serving many separate businesses, each with its data kept apart.

    Mining spare parts

    A global mining company held large stocks of spare parts and still ran out of the ones it needed. With a supply chain AI partner, we forecast each part at each site and measured the forecasts in money.

    Full story

    Planners could see the forecast, change it and record why.

  4. First steps

    1. DecisionWe pick one decision a forecast feeds today, and what a wrong call costs.
    2. DataYou share the history you have, however messy, and we tell you what it can support.
    3. PilotOur forecast runs beside yours for a few cycles. You switch only if ours is better.
    More detail

    Before building anything, we ask what decision the forecast is for. Sometimes the answer changes the job.

    We test every forecast against what you use now, on the same items and period. If ours isn't better, we'll tell you.

    For a small firm, that can be a forecast that updates the planning sheet you already use. For a large one, a platform serving many teams.

    Sometimes a forecast request turns out to be a planning problem. When it does, we say so early.

  5. Get in touch

    Tell us what you forecast now and what it costs when it's wrong. We'll tell you if a better forecast would help.

    Have a chat