How is the forecast built?

From your store's own orders: every day ahead is the same weekday a year earlier, moved forward and grown at the pace your orders are running at. Then the same engine as your P&L prices it, with your costs as they will be on that day.

The pace compares your most recent complete weeks, up to thirteen of them, with the same weeks a year earlier, and it is kept between a quarter of last year's level and three times it, so one odd month cannot run away with it. Each order moves with its lines, parcels and payments, and each refund moves with its own day, so the returns ahead look like your returns did. The reasoning behind each step is set out in forecasting sales from your own orders.

Pricing follows your rules exactly as the P&L does. Product costs, logistics and payment fee rules apply on their own dates, including a change you have saved with a future effective-from date, and overhead comes from your overhead book, to the day. Ad spend runs per platform: each at last year's spend grown by its own rate, or, where a platform barely ran a year ago, its last four weeks repeated, and a platform you paused stays paused. Marketing spend you entered on the Overhead page carries on as you set it up, and last year's Black Friday season repeats unless you switch it off.

A store with less than a year of orders has no season to read yet, so its forecast repeats the last four complete weeks, weekday by weekday, for the rest of this month and the next three, and the page says so. The forecast starts once a store has four weeks of orders.

This month mixes both: the days already gone are your actual figures and the rest are forecast. The forecast is rebuilt every day and after every sync, and it covers one store at a time; with several stores selected, it shows the store named in the page's header. How far ahead it can be trusted is set out in how accurate the forecast is.