What is marketing forecasting?
A forecast answers a simple question: if we carry on as planned, what happens next? It sets targets, sizes budgets, and gives the business an early warning when results start to run behind.
Marketing forecasts combine patterns in history, such as growth and seasonal peaks, with what the team already knows about the future: planned spend, promotions, launches, and holidays.
What are the main forecasting methods?
| Method | How it works | Good for |
|---|---|---|
| Exponential smoothing | Weights recent periods more heavily, with trend and seasonal parts | Stable series with clear seasons |
| ARIMA | Models a series from its own past values and past errors | Series with steady patterns over time |
| Regression with drivers | Predicts outcomes from inputs like spend, price, and promotions | Forecasts that respond to plans |
| Machine learning | Learns patterns across many series and drivers at once | Many products, regions, or channels |
| Mix-model forecasts | Uses a marketing mix model to forecast under a planned budget | Linking spend plans to sales |
How forecast accuracy is measured
| Measure | What it shows | Watch out for |
|---|---|---|
| MAPE | Average error as a percentage | Blows up when actual values are near zero |
| WAPE | Total error as a share of total actuals | Hides which periods were worst |
| Bias | Whether the forecast runs consistently high or low | Easy to miss if you only track error |
| Interval coverage | How often actuals land inside the forecast range | Ranges that are too narrow to be honest |
What is an example of marketing forecasting?
A holiday forecast that ran hot
A retailer's forecast for November said sales would grow 12%. By mid-month, actuals were tracking 4% behind.
A backtest showed the method had overestimated the last three holiday seasons too, because it treated an unusually strong year as the new normal. The team adjusted the holiday pattern, reported a range instead of a single number, and set a weekly check against actuals for the rest of the season.
Forecasting vs. planning vs. simulation
| Forecasting | Planning | Simulation | |
|---|---|---|---|
| Question | What will happen? | What will we do? | What would happen if we did something else? |
| Output | Expected results with a range | Budgets, targets, and a schedule | Results for each alternative |
| Updated | As actuals arrive | When decisions change | When a choice is being made |
Why does marketing forecasting matter?
Every target, budget, and inventory order rests on a forecast. When it's wrong, the business either overspends chasing a number that was never real, or runs out of stock when demand shows up.
- Earlier warnings. Comparing actuals with the forecast shows trouble weeks before the quarter closes.
- Honest targets. Ranges show how ambitious a goal really is.
- Better methods over time. Tracking error and bias by period shows which forecasts to trust.
How does Alli do this?
Alli by PMG's Media Planner moves plans into one governed workspace where teams build plans around business goals and simulate budgets against historical benchmarks. Planned spend sits next to actual performance by channel, market, and time period, so pacing and variance are visible without manual reconciliation.
For budget forecasts, Alli Scenario Planner sets a forecast period, channels, and budget, and shows predicted revenue with recommended spend for every channel, blending a media response model built on historical sales with current brand signals.
Explore Alli Media PlannerMarketing forecasting FAQ
What is the best method for marketing forecasting?
There isn't one. Simple methods like exponential smoothing often do well on stable data, and driver-based models do better when plans change. Backtest a few on your own history and keep the one that performs best.
What is MAPE?
Mean absolute percentage error: the average size of forecast errors as a percentage of actual values. Lower is better, but it can mislead when actual values are small.
What is forecast bias?
A forecast's tendency to run consistently high or low. A biased forecast can have modest average error and still mislead every plan built on it.
How often should a marketing forecast be updated?
Whenever new actuals arrive that change the picture, and on a regular rhythm such as weekly or monthly. Compare each update with the last to see how the forecast is moving.
Last updated Back to the glossary