Revision sheet · Unit 4: Marketing
4.3 Sales forecasting
HL This subtopic is HL only. If you are taking SL, it is extension material.
Key points
- A sales forecast estimates future sales, in units or value, over a stated period; budgets, cash flow forecasts and production and workforce plans start from it.
- Past sales mix a trend, seasonal variation, cyclical variation and random variation; extrapolating the trend and adjusting for the season gives a basic forecast.
- Simple linear regression (HL toolkit): plot x and y on a scatter diagram, fit a line of best fit y = a + bx, and judge the correlation, from −1 to +1.
- Interpolation reads the line inside the data; extrapolation goes beyond it and is less reliable. Correlation is not causation.
- Qualitative methods, such as expert opinion, sales force estimates and surveys, are needed when there are no useful past data, but are open to bias; test marketing scales up real sales from a trial area.
- Benefits include better planning of production, stock, staff and cash, support for raising finance, targets for control, and coordination between departments.
- Limitations include the assumption that the past continues, external shocks, poor or missing data, method weaknesses, bias, cost, and falling accuracy the further ahead the forecast looks.
- Use forecasts well by combining methods, forecasting a range, updating regularly, measuring past accuracy and building flexibility into plans.
- Plan with a range of low, expected and high cases, combined where possible into an expected value (the sum of probability × outcome), read beside the worst case and tested for its sensitivity to the probabilities.
- Measure past forecasts with the forecast error, (actual − forecast) ÷ forecast × 100. The mean of the signed errors shows bias; their spread shows how wide a range the next forecast needs.
Common examination mistakes
-
“A strong correlation proves that one thing causes the other.”
Instead: Both may be driven by a third factor, or the link may be coincidence; causation needs further evidence.
-
“The expected value must be one of the possible outcomes.”
Instead: It is a weighted average, so it is often a figure that cannot actually happen.
-
“Cyclical variation and seasonal variation are the same thing.”
Instead: Seasonal variation repeats at known times within each year; cyclical variation follows the economy over several years, with no fixed timing.
-
Reading correlation as causation.
Instead: Name a possible third factor and say how the business could test the link, such as a controlled trial.
Key definitions
The IB glossary terms of 4.3, in the platform's own words. Terms beyond the course are not included.
- Correlation
- The extent to which two variables are linearly related: with positive correlation they tend to rise together, with negative correlation one tends to fall as the other rises, and with no correlation there is no straight-line pattern.
- Cyclical variation
- Rises and falls in sales that follow the business cycle, the economy's movement between boom and recession, over several years.
- Expected value
- The average outcome of an option over many repetitions, found by multiplying each outcome's payoff by its probability and adding the results, then subtracting the option's cost.
- Extrapolation
- Projecting a trend or a line of best fit beyond the data already collected, for example into future periods, to produce a forecast.
- Line of best fit
- The straight line that most closely follows the pattern of points on a scatter diagram, with the points balanced as evenly as possible above and below it.
- Qualitative sales forecasting
- Forecasting sales from opinions and judgement rather than from past figures, for example from experts, the sales team or surveys of customers' buying intentions.
- Quantitative sales forecasting
- Forecasting sales from numerical data, usually the business's own past sales or figures for something its sales depend on, such as advertising spending or the weather, by finding a pattern and projecting it forward.
- Random variation
- Unpredictable changes in sales caused by one-off events, such as extreme weather, a strike or a sudden news story, that follow no pattern.
- Sales forecast
- An estimate of a business's future sales, in units or in value, over a stated period, based on past data, research or judgement.
- Scatter diagram
- A graph that plots pairs of values for two variables as points, with the independent variable on the horizontal (x) axis and the dependent variable on the vertical (y) axis, to show whether they are related.
- Seasonal variation
- A regular, repeated rise and fall in sales at particular times of the year, week or day, around the trend.
- Simple linear regression
- A statistical method that fits a straight line to data on two variables, an independent variable (x) and a dependent variable (y), to describe how y tends to change as x changes and to make forecasts.
- Test marketing
- Launching a new product, or a change to the marketing mix, in a limited area or to a small group of customers before a full launch, to see how buyers respond.
- Trend
- The underlying long-term direction of a set of figures over time, such as sales rising or falling, once short-term ups and downs are set aside.
Formulae
Each with what it measures. The formula sheet shows which ones the examination provides.
| Formula | What it measures |
|---|---|
| Average change per period = (latest value − earliest value) ÷ number of periods between them | The typical rise or fall from one period to the next across a run of data, used to extend a trend forward. A simple technique, not one the guide names. |
| Seasonal share = sales in the period ÷ sales in the year × 100 | How much of a year's sales one period accounts for, which shows how strongly sales follow the seasons. |
| Line of best fit y = a + bx | The straight line that most closely follows the pattern of points on a scatter diagram, with the points balanced as evenly as possible above and below it. |
| Forecast error = (actual − forecast) ÷ forecast × 100 | How far the actual result differed from the forecast, as a percentage of the forecast: a positive figure means the result beat the forecast, a negative one that it fell short. |
| Expected value = the sum of (probability of each outcome × value of that outcome) | The average outcome of an option over many repetitions, found by multiplying each outcome's payoff by its probability and adding the results, then subtracting the option's cost. |
Frameworks and tools
From the frameworks library: what each is for, a situation it suits, and one limitation.
-
Simple linear regression IB toolkit
- Purpose
- To test whether two business variables are linked and to make cautious forecasts, such as sales from advertising spend or demand from temperature.
- Use it when
- Testing whether two business variables move together, such as advertising and sales, or temperature and ice cream demand, before planning around the link.
- One limitation
- Correlation does not show causation; a third factor may drive both variables.
Apply it
Short exercises on fictional businesses, written for the platform; none is an IB question. Answer first, then open the outline of what a strong answer does.
- Ferrymoor Bikes forecasts next year's sales from five years of monthly sales, which show a steady upward trend and a peak every spring.
Explain two benefits of the sales forecast for Ferrymoor Bikes.
Show a model answer outlineModel answer outline
- Operations: it can plan production and stock for the spring peak, avoiding lost sales and unsold bikes.
- Human resources: it can hire seasonal staff in time.
- Finance: it can plan the cash needed to build stock before the peak, and set budgets and targets.
- Explains each benefit with Ferrymoor Bikes' trend and spring peak.
- Ferrymoor Bikes' managers want to rely on past sales alone to plan the next three years.
Discuss the limitations of using past sales to forecast Ferrymoor Bikes' sales.
Show a model answer outlineModel answer outline
- The forecast assumes the past continues, but the economy, cycling fashions, a new rival or a run of bad weather could change the trend.
- Random variation and poor data make any forecast uncertain, and accuracy falls the further ahead it looks.
- Qualitative methods, such as sales force estimates or surveys, can add what the data miss, though they bring their own bias.
- Judges how far to rely on it, for example to use a range of forecasts and update them every quarter.
- Ferrymoor Bikes finds a strong positive correlation between its monthly sales and the hours of sunshine each month.
Explain why Ferrymoor Bikes should be careful about using this correlation to forecast sales.
Show a model answer outlineModel answer outline
- Correlation is not causation: sunshine and sales both rise in spring and summer, with holidays and longer evenings, so other factors may drive both.
- Sunshine itself cannot be forecast far ahead, so it is a poor basis for a long-range plan.
- The relationship may change, and extrapolating beyond the data is less reliable than reading within it.
Answer from memory
15 recall questions from the 4.3 notes
Answer each question from memory, out loud or on paper, before you reveal the answer. Your ratings also schedule these questions in your flashcards.
-
What is a sales forecast?
An estimate of the sales a business expects to make over a stated future period, in units or in value, based on past data, research or judgement.
How well did you recall it? -
Name the four components of a series of sales over time.
Trend, seasonal variation, cyclical variation and random variation.
How well did you recall it? -
What is the difference between seasonal and cyclical variation?
Seasonal variation repeats within each year, week or day; cyclical variation follows the economy over several years.
How well did you recall it? -
What is extrapolation?
Extending a trend or line of best fit beyond the data already collected, for example into the future, to make a forecast.
How well did you recall it? -
In the equation y = a + bx, what do a and b represent?
a is the intercept, the value of y when x is zero; b is the gradient, the change in y for each one-unit change in x.
How well did you recall it? -
The line of best fit is y = 50 + 3x. Forecast y when x = 20.
y = 50 + 3 × 20 = 110.
How well did you recall it? -
What does a correlation coefficient of 0.97 show?
A strong positive correlation, with the points lying close to a rising straight line.
How well did you recall it? -
Why does a strong correlation not prove that x causes y?
The two may move together because a third factor drives both, or by coincidence, so there must be a business reason for believing one affects the other.
How well did you recall it? -
When is qualitative forecasting most useful?
When there are no useful past data, as with a new product or market, or when the future is expected to differ from the past.
How well did you recall it? -
Give three benefits of sales forecasting.
For example, planning production and stock, planning staff, building cash flow forecasts and budgets, supporting a case for finance, and setting targets.
How well did you recall it? -
Give three limitations of sales forecasting.
For example, the future may not follow the past, external shocks, poor or missing data, bias in judgement, cost and time, and lower accuracy further ahead.
How well did you recall it? -
Why might a business forecast a range rather than a single figure?
Because a range shows the uncertainty, so managers can see the risk, test their plans against the low case and prepare responses in advance.
How well did you recall it? -
How is the expected value of an option calculated?
Multiply each outcome by its probability and add the results; subtract the option's cost if it has one.
How well did you recall it? -
Forecasts were 100 and 120; actual sales were 92 and 126. What were the forecast errors, and what does their mean show?
(92 − 100) ÷ 100 × 100 = −8% and (126 − 120) ÷ 120 × 100 = +5%. Their mean is −1.5%, so on average the forecasts were slightly too high, but the size of the errors matters more than the lean.
How well did you recall it? -
Why might a business plan for more than its expected sales?
When a lost sale costs more than an unsold unit, for example when the contribution per unit is large and unsold stock can be sold off, erring on the high side costs less than running out.
How well did you recall it?
Arjun Agrawal | The Business Brief. Gathered from the platform's notes, glossary, calculation register and frameworks library. Original teaching material; not an IB publication.