What assumptions is your retail demand forecast built on?
The short answer
Ask for six things behind any retail demand forecast: the sales history window, the forecasting method, how promotions were treated, whether stock-outs were corrected, which customers drive the volume, and the error range from the last back-test. If a forecast cannot show all six, hold off on buying against it until it can.
Why it matters
A forecast with hidden assumptions becomes a buy you cannot defend. If last year's stock-outs went uncorrected, the model understates demand and you lose full-price sales. If a one-off promotion stayed in the baseline, you overbuy and pay for it in markdowns and held inventory.
The question to ask
Copy this into Lexi as written.
What the answer looks like
Illustrative output on sample data
Illustrative example: how Lexi answers this for an outdoor retailer with 38 stores. The spring forecast of 4,200 rain jackets uses 104 weeks of store and online sales, matched to one record per customer.
- Removes last March's 30% off event, about 600 units that will not repeat.
- Adds back demand lost in weeks when core sizes were out of stock.
- Customer mix: 55% repeat buyers, 30% new and 15% promotion-led.
- Back-tested error: plus or minus 12%, a range of 3,700 to 4,700.
What the data shows
Repeat buyers carry most of the forecast, and their demand has held steady for two seasons.
What it means
The promotion-led 15% is the soft spot. Without a spring sale, many of them may not buy.
What to do
Commit depth to the roughly 3,600 units that repeat and new buyers support, and keep an open order for the rest.
The range comes from last season's back-test, so a new style or a late season can push results outside it. Would you like the forecast split by store?
Frequently asked questions
Why do forecasts underestimate demand after a stock-out?
Sales history records what you sold, and that is capped by what you had on the shelf. If a size sold out for three weeks, those weeks show low sales that look like low demand. Unless lost demand is estimated and added back, the next forecast repeats the shortfall and the next buy comes in too shallow.
What is explainable demand forecasting?
Explainable demand forecasting means every forecast arrives with the inputs, method and adjustments that produced it, in terms a planner can check. You can see which sales weeks fed the model, what was removed as a one-off and how much uncertainty surrounds the number. Without that record, nobody can tell a good forecast from a lucky one.
Which demand forecasting methods suit a mid-market retailer?
A seasonal baseline built on at least two years of weekly sales, adjusted for promotions and stock-outs, is a sound starting point for most ranges. Machine learning models can add accuracy across large ranges, but they are harder to explain to a buying team. Choose the simplest method whose back-tested error you can accept, and write it down.