A seasonal sales increase does not automatically justify multiplying every pharmacy order. The apparent change might reflect extra trading days, a one-off bulk purchase, a stockout last year or a promotion that will not happen again.
Seasonal demand forecasting is a planning exercise for the products your pharmacy is authorised and professionally prepared to supply. It is not a recommendation to promote clinically unnecessary medicine use.
Start with a comparable baseline
Choose a product group and a planning period. Record actual trading days, sales units, unavailable-item observations and exceptional orders. Compare equivalent periods rather than assuming that identical calendar dates describe identical demand.
In a hypothetical example, last year's event week supplied 140 units over seven trading days. An ordinary week supplied 100 over five days. Both average 20 units per trading day; the larger event total alone does not demonstrate stronger daily demand.
Decide whether your model uses trading or calendar days and keep the units consistent with supplier lead times. Retain the assumption rather than hiding it inside a spreadsheet formula.
Separate demand from availability
A sold-out line can show low recorded sales during high demand. Review stockout periods and unavailable-item logs before treating observed sales as the entire demand signal. Do not invent an exact lost-sales figure when the records do not support one.
Segment exceptional bulk orders and explain whether they are expected to recur. Review range changes, price changes and changes in service hours. A product introduced halfway through the comparison period has less observed history than a long-established line.
Build a small scenario range
Suppose a defensible baseline is 20 units per trading day and the upcoming period has six trading days. Base demand is 120 units. For planning discussion, you might examine 96 and 144 as illustrative lower and higher scenarios—not statistically validated confidence limits.
Compare each scenario with usable stock, confirmed incoming quantities, expiry exposure and available purchasing cash. A demand forecast does not by itself determine an order quantity. The stock position and supplier pack sizes still matter.
Record the assumptions behind any event uplift. The owner should be able to challenge “increase 30%” by asking where the figure came from, what changed and how the result will be reviewed.
Backtest against a simple alternative
Before adopting a complicated model, test it on a previous period whose actual outcome is known. Compare the forecast with a simple baseline such as the comparable-period average.
For four hypothetical periods, absolute errors of five, eight, four and seven units average six. Calculate the same measure for the baseline. If the elaborate method averages ten units of error, it has not improved that comparison merely because the chart looks sophisticated.
Watch for systematic overprediction and underprediction as well as average absolute error. Avoid percentage-error measures that divide by zero demand. Small samples require judgement and should not be presented as proof of future accuracy.
Close the learning loop
After the event, record actual demand indicators, stockouts, leftover usable stock and expiry exposure. Explain where the forecast was wrong and change one assumption at a time. Keep the prior forecast so hindsight does not replace the original decision.
Software teams should version forecasts and preserve source periods. Reviewers need to distinguish original expectations from a report recalculated after the outcome became known.
Use RunMyPharmacy's reporting workflow and purchase-order records to evaluate the data available for this worksheet. Confirm export coverage; this article does not claim built-in predictive forecasting or guaranteed demand accuracy.
Disclosure: created for the RunMyPharmacy team with AI drafting assistance and checks of examples, sources and product links. We build the linked product. Numerical examples are hypothetical.
Top comments (0)