How Much Advertising Moves Sales

ConversionAssmus, G.; Farley, J. U.; Lehmann, D. R. (1984) · 1984Journal of Marketing Research
Topicsadvertising elasticity·meta-analysis·carryover·replication analysis·econometrics·cpg

Your measurement team reports a high advertising sales elasticity, the percentage change in sales for each 1% change in advertising. A media plan depends on it. Do you approve the plan, or first ask how the model handled carryover (the effect that keeps working after spending stops) and other sales controls?

Estimates of how much advertising moves sales vary with how the model is built.

Data chart

Models that leave out carryover or other sales controls report a larger same-period sales response

Extra percent change in same-period sales for each 1% more advertising that a model reports when it leaves the factor out, compared with models that include it.

Model leaves out carryover+0.336Model leaves out other salescontrols+0.103
  • Leaving out carryover moves its effect into the same-period estimate.

Models that leave out carryover or other sales controls report higher short-term elasticity.

Differences in short-term elasticity between models without and with each factor, after allowing for the other differences between studies. Other sales controls are economic and household measures, such as income or family size, that also move sales. Based on 128 models across 22 studies of predominantly mature, frequently purchased U.S. consumer products through 1980; exact values +0.336 and +0.103.

Data chart

Across published models, sales rise in the same period and about half the effect carries forward

Averages across published models: percent change in same-period sales for each 1% more advertising, and share of each period's advertising effect that carries into the next period, in percent.

0.22%Same-period sales change for each 1percent more advertising47%Share of each period's effect carriedinto the next period
  • Test new models against these averages, not against zero.

Across published models, sales rise about 0.22% in the same period for each 1% more advertising, and about 47% of each period's effect carries into the next period.

Simple averages, not adjusted for differences between studies, of 128 short-term elasticities and 114 carryover coefficients from 22 published studies of predominantly mature, frequently purchased U.S. consumer products through 1980; exact values 0.221 and 0.468, which the paper rounds to about 0.3 and 0.5 as starting values for testing a new model.

Action guide

  1. Ask how the model was built before approving the plan.Does it include carryover and other sales controls? Does it pool data?
  2. Treat estimates from models missing carryover or other sales controls as likely overstated.Ask for a model that includes them.
  3. Use published averages only as a check.Apply them only to mature, frequently purchased U.S. consumer products, allowing for model differences.
  4. Account for model differences before comparing markets.A gap that remains, such as Europe's higher estimates, may be real.

Evidence

  • Models without a carryover term report short-term elasticity about 0.34 higher than models with one.
  • Models that omit other sales controls report short-term elasticity about 0.10 higher than models that include them, and may overstate long-term effects even more.
  • Models pooling data across regions or customer groups report short-term elasticity about 0.18 higher than models using one time series, and fit sales better.
  • The simple average short-term elasticity is about 0.22. Adjusted for differences in models, data and settings, it is about 0.70, so advertising may work better in the short run than 0.22 suggests.
  • In models with carryover, about 47% of each period's sales effect carries into the next period.
  • After accounting for model differences, short-term elasticity differs little across product categories but is about 0.10 higher for advertised food products. It is higher in Europe than in the U.S., which may mean European firms underspend because of media restrictions, or U.S. firms overspend.

Key takeaway

Check how a model is built before trusting its advertising sales elasticity.

Limitations

The studies cover mostly mature, frequently purchased U.S. consumer products through 1980 and measure sales, not profit. New products, durable goods, services, other regions and digital media get little or no coverage.

Source

Assmus, G., Farley, J. U., & Lehmann, D. R. (1984). How advertising affects sales: Meta-analysis of econometric results. Journal of Marketing Research, 21(1), 65–74. https://doi.org/10.1177/002224378402100107

Exact values behind the rounded figures: short-term elasticity 0.336 higher in models without a carryover term, 0.103 higher in models without other sales controls, 0.176 higher with pooled data and 0.101 higher for advertised food products; a simple average short-term elasticity of 0.221 and an adjusted average of 0.695; an average carryover coefficient of 0.468, printed as 47%.

Read the paper ↗