Context
Online marketplace with ~1,100 orders and ~1,650 items sold per day across desktop, mobile web and two native apps. Order-line data, Aug 1–5, 2022.
The problem
Forecast end-of-month revenue and orders from only a 5-day trend, so inventory and logistics can plan ahead.
Technical approach
A daily-growth forecast: run-rate over 5 days compounded at an adjustable daily growth rate (default 2%), by device.
Business value
Cumulative unique orders by day
Aug 1–5, with linear trend. Follows the device filter.
Forecast to end of August
Actuals (Aug 1–5) then forecast (Aug 6–31)
Device performance
Share of orders vs share of revenue, Aug 1–5 (all devices)
Daily growth trend
Day-over-day change. Follows the device filter.
Top insights
Recommended actions
What I would do with these findings, in priority order. Figures are for all devices.
Method. Baseline = average daily value over Aug 1–5. Each forecast day = baseline × (1 + g)n, n = days after Aug 5, g = assumed daily growth. Observed growth over five days is noisy, so the rate is left adjustable. Revenue is line-level revenue after discounts, in the source system’s currency units. Rows from earlier months in the extract are excluded.