The HOOF approach to demand forecasting

A market-size estimate is a snapshot. Strategy also needs a view of direction and speed. HOOF translates historic demand and its drivers into a reasoned forecast for each product-market segment.

When to use it

• Use the process whenever a decision depends on future market demand.

Origins

I developed HOOF as a four-stage forecasting discipline: Historic growth, Origins of past growth, Outlook for those drivers, Forecast growth. A strict acronym would be HDDF; HOOF is easier to remember and suggests the rising path every forecaster would like demand to follow. The football metaphor is deliberately imperfect: like a kicked ball or a product life cycle, growth can reach a summit and fall; see The strategic condition matrix (Arthur D. Little).

What it is

Growing markets usually give a company better odds than shrinking ones, but extrapolation is not enough. HOOF links a measured trend to the forces that produced it, asks how those forces will change and then derives the forecast. Its value is the explicit chain of reasoning, which can be challenged and updated.

How to use it

Apply the stages separately to each important segment and keep real volume growth distinct from nominal revenue growth.

Historic growth

Assemble a recent demand series from market research or a carefully constructed estimate. Do not anchor on the latest observation. A market that rose 8 per cent last year may not have an 8 per cent trend: it might have fallen two years ago, remained flat and then rebounded by 8 per cent, leaving average annual growth near 2 per cent.

Prefer a compound rate over several recent years and inspect the path. If annual movements were unusually volatile—as in 2008–11—use smoothing such as a three-year moving average before estimating trend. Measure comparative market growth in real terms. Nominal growth includes price change; real growth removes the relevant market-price deflator and better approximates volume. Add price forecasts back when converting demand into revenue and a financial plan.

Drivers past

Identify the forces behind the observed movement. Candidates include per-capita income, total or segment-specific population, government policy and purchasing, awareness and promotion, outsourcing or other structural shifts, price, fashion, weather and climate.

Look for derived demand. Steel depends partly on automobiles, ships, capital equipment and construction. A Tier 1 seat supplier follows vehicle production, which affects a Tier 2 mechanism supplier and its specialist-steel provider. Yet derived demand is never the whole explanation: changing seats per car, 4x4 popularity or a new seat technology can break a simple relationship. Hotel demand may follow coach tourism while also responding to exchange rates and competing destinations.

Drivers future

Assess the future direction, magnitude and timing of every important driver. Will demographic growth persist? Could regulation, taxation or fashion change? What is the outlook for customer and complementary sectors?

Pay particular attention to the economic cycle and demand elasticity. A downturn matters greatly to cyclical markets and less to relatively inelastic essentials; see Income elasticity of demand. Build alternative cases where a driver is uncertain rather than burying uncertainty in one average.

Forecast growth

Combine the historic trend with the expected change in drivers. Record evidence and judgement separately, show the contribution of each driver and reconcile the result with external constraints.

For a service to older people: Step 1 identifies historic growth of 5–10 per cent; Step 2 attributes it to income, growth in the elderly population and awareness; Step 3 expects income to continue, the population effect to strengthen and awareness to spread; Step 4 concludes that future growth could exceed 10 per cent. Display plus, minus and neutral effects so readers can see why the forecast differs from history.

The HOOF approach to demand forecasting: an example

A crane-market example shows the danger of exclusion. With no direct series available, a planner equated crane demand to an OECD forecast of 2.4 per cent engineering growth. That ignored destocking, the second-hand market and an approaching downturn in high-rise construction. Including those drivers changed a second 2.4 per cent growth assumption into a possible decline of 10 per cent a year for two or three years.

The lesson is to include material drivers even when only structured qualitative evidence exists. False precision from one available statistic is weaker than a transparent synthesis of quantitative and well-sourced qualitative evidence.

Top practical tip

Apply all four HOOF stages to every major segment and make the contribution of each demand driver explicit.

Top pitfall

Do not exclude a material driver simply because hard data are unavailable; document the qualitative evidence, uncertainty and judgement instead.

Further reading

Hyndman, R.J. and Athanasopoulos, G. (two thousand and twenty-one) Forecasting: Principles and Practice, third edition. Melbourne: OTexts.

Lawrence, M., Goodwin, P., O’Connor, M. and Önkal, D. (two thousand and six) “Judgmental Forecasting: A Review of Progress Over the Last Twenty-Five Years.” International Journal of Forecasting.