Warehouse forecast: predict outbound, labor, inbound and capacity

Four plain-language forecasts built from your own history: how many orders are coming, how many people you need, how many trucks arrive, and when you will run out of space. Every number explains how it was calculated.

What it answers#

• Outbound: how many orders, lines and units will ship each day for the next 14 days?
• Labor: how many pick / putaway / pack hours and suggested headcount per day?
• Inbound: how many trucks, pallets and units arrive each day for the next 14 days?
• Capacity: what will location occupancy look like over the next 30 days, and when does it cross 85% / 95%?

How each number is calculated (plain language)#

Outbound volume#

We take the average outbound volume of the same weekday over the past 4 weeks (warehouses have a strong day-of-week pattern — Mondays restock, Fridays rush), then multiply by a trend factor. The trend factor is the recent 4-week daily average divided by the previous 4-week average, clamped between 0.7 and 1.5 so one crazy week cannot blow up the forecast. Draft and cancelled orders are excluded.

Labor demand#

UPH (units per hour) comes from your real history: total units handled divided by total completed task hours, calculated separately for pick, putaway and pack. Tasks shorter than 1 minute or longer than 8 hours are treated as outliers and removed. Forecast units ÷ UPH = hours needed; hours ÷ shift length (default 8h, adjustable) = suggested headcount, rounded up. If your warehouse has no pack tasks, pack hours show 0.

Inbound volume#

Planned receipts grouped by expected date for the next 14 days. Unit counts are adjusted by each vendor’s historical delivered-vs-planned ratio over the last 90 days (if a vendor usually delivers 90% of what was planned, we forecast 90%). Truck count follows the plan as-is. Pallets come from the receipt plan, or are estimated from units when the plan has no pallet count.

Capacity#

Current occupied locations + inbound pallets (forecast) − outbound pallets (forecast units ÷ average units per pallet). The yellow line is 85% occupancy, the red line 95%. Beyond the 14-day inbound forecast, inbound uses the 14-day daily average. Cart-type locations are excluded from the total.

Reading the accuracy badge#

Every night the system compares last week’s predictions against what actually happened and shows "last week’s forecast was X% accurate". It is computed as 1 − MAPE (mean absolute percentage error) on daily unit counts. When there is not enough history yet, it shows "collecting data" — that is honest, not broken. If accuracy drops, check for one-off events (a big promotion, a carrier strike) that the moving average cannot see.

Drill down: click any day#

• Outbound card → click a day: the shipments created that day (number, status, lines).
• Inbound card → click a day: receipts expected that day (receipt no, vendor, status).
• Labor card → click a bar: that day’s tasks of that type (task no, assignee, status, duration).
• Capacity card → click a day: that day’s projected occupancy plus current zone occupancy breakdown.

When it runs#

The forecast batch runs automatically every day at 03:00 PDT and writes the next 14 / 30 days into the cache — the page always loads instantly. Use "Recalculate" on the page to refresh immediately after changing parameters (moving-average weeks, shift length); new parameters apply to the next run.

FAQ#

Q: Why is my forecast all zeros? A: The warehouse has less than a week of history, or all recent shipments are drafts. Create real outbound orders (or wait for the nightly batch after real activity) and it will fill in.

Q: Does forecasting change my operations? A: No. It only reads history — it never creates, modifies or deletes orders, tasks or inventory.