American Hoist made sales judgment anonymous, iterative and accurate

American Hoist and Derrick faced a familiar forecasting problem: historical sales data alone were not enough when market assumptions were changing. But simply asking managers for opinions risked hierarchy, anchoring and loud-voice bias.

The company applied the Delphi method to sales forecasting. Managers gave judgments through a structured process, received feedback about the group view, and revised their estimates without turning the forecast into a meeting-room contest.

The neat move was procedural rather than computational. American Hoist did not pretend judgment could be removed; it designed a mechanism that made judgment more disciplined and less political.

After one year of use, the published case reported that 1975 sales forecast error fell below 1%, far better than the previous 10 years.

Structure expert judgment before averaging it

The Delphi process kept expert knowledge while reducing pressure to agree with senior voices.

Iteration let managers update assumptions after seeing the anonymous group signal.

The method worked where pure extrapolation failed because the underlying market conditions were shifting.

Expert judgment is not the enemy of analytics. The trick is to make it explicit, independent, and revisable before it enters the forecast.

The Interfaces article presented the method as a practical way to incorporate management judgment into sales forecasts. It reported less than 1% forecast error for 1975 after one year of Delphi use.