How to forecast demand for your Shopee products

Every stock decision is really a bet on the future: you order today for sales that will happen tomorrow, next week, next month. Demand forecasting is simply making that bet informed rather than blind — using what you know to predict how much you will sell, so you can stock the right amount. It sounds like an advanced, data-scientist discipline, but for a Shopee seller it is mostly the disciplined use of your own sales history plus awareness of what is coming. You do not need a crystal ball; you need to stop ignoring the signals you already have.

Good forecasting is what lets you hold the right amount of stock — enough to avoid stockouts without overstocking. This guide explains how to forecast demand well enough to stock right, without overcomplicating it. As always, the specifics depend on your business; this is an educational overview.

Your sales history is the foundation

The single best predictor of future demand is your own past demand. Before any clever technique, the foundation of forecasting is simply looking at how much each product has actually sold over recent periods and using that as your baseline expectation. A product that has steadily sold five a day is, absent other information, likely to keep selling around five a day — so that is your starting forecast.

This sounds obvious, yet many sellers do not do it. They order by feel, by enthusiasm for a product, or by whatever they happened to sell last, rather than by a clear look at each product's actual sales rate over time. The discipline of forecasting starts here: know your sales velocity per product, based on real history, and treat it as your baseline. Everything else — seasonality, promotions, trends — is an adjustment to this baseline, not a replacement for it. Get the baseline right by actually reading your sales history, and you are already forecasting better than most sellers. This is one more reason accurate sales records pay off: they are the raw material of every forecast.

Adjust for what you know is coming

A baseline from history assumes the future looks like the past, which is usually true — except when you know it will not. The second layer of forecasting is adjusting your baseline for foreseeable changes:

• Seasonality. If your products sell more in certain seasons, holidays or times of month, your forecast for those periods should rise above baseline. History helps here too — last year's seasonal pattern informs this year's.
• Promotions and campaigns. A promotion or marketplace campaign predictably lifts demand, sometimes dramatically. Forecasting for a promoted period means stocking for the spike, not the baseline, or the promotion's success becomes a stockout.
• Trends. A product gaining or losing popularity is drifting away from its historical baseline; adjust for the direction of travel rather than assuming the past rate holds.
• Known events. Anything you can see coming — a new competitor, a supplier delay, a marketing push — that will move demand belongs in the forecast.

The skill is not predicting the unpredictable but incorporating the predictable — the things you already know are coming but might otherwise forget to factor in. A promotion you planned should never cause a "surprise" stockout, because it was never a surprise; it was foreseeable demand you failed to forecast. This is where many stockouts come from: not unknowable events, but knowable ones the seller did not stock for.

Respect uncertainty with buffers

Forecasting is prediction, and predictions are never exact — so a good forecast is not a single number but a number plus an acknowledgement of how uncertain it is. This is where the safety buffer comes in: you stock for your forecast, plus extra to cover the chance that real demand runs higher.

How big a buffer depends on how uncertain the forecast is. A stable product with steady history and no upcoming changes has a reliable forecast, so it needs little buffer. A product with volatile sales, an upcoming promotion of uncertain impact, or a short unpredictable history has a shaky forecast, so it needs more buffer to protect against the wide range of possible outcomes. Matching your buffer to your uncertainty is what keeps forecasting honest — it admits that you are estimating, and protects you against being wrong in the costly direction. The profit calculator helps you weigh this: a high-margin product's stockout is expensive, so it may justify a larger buffer against forecast error.

How to forecast in practice

A practical, non-overwhelming forecasting routine:

1. Establish each product's baseline from history. Look at recent actual sales per product and use that rate as your default expectation. This alone is a big step up from ordering by feel.
2. Adjust for the foreseeable. Raise or lower the baseline for known seasonality, planned promotions, visible trends and upcoming events. Stock for the spike when a spike is coming.
3. Add a buffer sized to uncertainty. Stock for your forecast plus safety stock, larger where the forecast is shaky and smaller where it is reliable.
4. Review and learn. Compare what you forecast to what actually sold, and let the misses sharpen your next forecast. Forecasting improves with feedback.

Do these and your stock decisions become informed bets rather than blind ones. You will not be right every time — no one is — but you will be roughly right far more often, which is what stocking the right amount requires.

The promotion that emptied its own shelves

A seller plans a big promotion on a product that normally sells five a day. They order their usual amount — a week or so of stock at the normal rate — because that is what they always order. The promotion launches, demand triples as intended, and the product sells out in two days. The promotion, which should have been a triumph, becomes a stockout: they lose the surge of sales they had specifically created demand for, and the product loses momentum right when it had the most.

The painful part is that this was entirely foreseeable. The seller planned the promotion; they knew demand would spike; and yet they stocked for the baseline as if nothing were coming. The forecast failure was not predicting an unknowable event — it was failing to incorporate a known one. Had they forecast for the promoted period — baseline times the expected promotional lift, plus a buffer for uncertainty about exactly how big the lift would be — they would have stocked for the spike and captured it. Instead, their own promotion emptied their shelves. That gap between "I knew demand was coming" and "I stocked as if it wasn't" is exactly what demand forecasting exists to close.

Common questions

How do I forecast demand for my Shopee products?

Start with your own sales history — the single best predictor of future demand. Look at how much each product has actually sold over recent periods and use that rate as your baseline expectation; a product that has steadily sold five a day will likely keep selling around that, absent other information. Then adjust the baseline for what you know is coming: seasonality (raise forecasts for busy seasons, informed by last year's pattern), planned promotions (stock for the spike, not the baseline), visible trends (a product gaining or losing popularity is drifting from its history), and known events like competitor moves or supplier changes. Finally, add a safety buffer sized to how uncertain the forecast is. This is not advanced data science — it is disciplined use of the signals you already have. The key skill is incorporating the predictable things you might otherwise forget to factor in, which is where most stockouts actually come from.

Why do my promotions cause stockouts?

Almost always because you stocked for your baseline demand rather than for the promotional spike you knew was coming. A promotion predictably lifts demand, sometimes dramatically, so forecasting for a promoted period means stocking for the surge, not the normal rate — otherwise the promotion's success empties your shelves and you lose the very sales you created demand for, plus the momentum the product had at its peak. The frustrating thing is that this is entirely foreseeable: you planned the promotion, so the demand was never a surprise, only an unforecast known event. The fix is to build promotions into your forecast explicitly — take your baseline, multiply by the expected promotional lift, and add a buffer for uncertainty about exactly how large the lift will be. Then you stock for the spike and capture it, rather than letting your own promotion cause a stockout right when demand is highest.

Do I need special software to forecast demand?

Not to start — good demand forecasting for most Shopee sellers is mostly disciplined use of your own sales history plus awareness of what is coming, which you can do without sophisticated tools. The foundation is simply reading each product's actual sales rate and using it as a baseline, then adjusting for foreseeable seasonality, promotions and trends, and adding a buffer for uncertainty. What you do need is accurate sales records to forecast from, since the whole method rests on knowing your real sales history per product — which is one more reason good bookkeeping and inventory tracking pay off. As your range and volume grow, software that surfaces sales velocity, seasonality and stock levels automatically makes forecasting faster and more accurate, especially across many products. But the discipline matters more than the tooling: a seller who reads their history and stocks for known events forecasts far better than one who orders by feel, regardless of software.

Informed bets beat blind ones

Demand forecasting is not fortune-telling; it is making your stock bets informed rather than blind. Build each product's baseline from its real sales history, adjust for the foreseeable — seasonality, promotions, trends, known events — and add a buffer sized to your uncertainty. The key skill is incorporating the predictable things you already know are coming, because most stockouts are not unknowable surprises but foreseeable demand the seller failed to stock for. Do this and you will be roughly right far more often, which is exactly what stocking the right amount requires.

Surfacing each product's real sales velocity and seasonality — the raw material of every forecast — is part of the operational picture SmartB Studio brings together for Shopee sellers, alongside reconciliation aiming for 98% automation, high by design and never total. See how it works, or start with the profit calculator.

Related: how much stock should a Shopee seller hold and stockouts on Shopee: the real cost of running out.