8.7
Split-Cable Experiments
Randomised in-market tests that deliver different advertising weight or copy to matched halves of a cable television audience and read the sales difference from the same households' purchase panel.
IRI's BehaviorScan; meta-analysed by Lodish et al., JMR (1995)
What it does
The cleanest causal read on television advertising there has ever been: same households, same stores, randomised assignment, purchases measured rather than recalled. Every geo-lift and incrementality test in this category is a descendant, and the effect sizes the category quotes exist because these experiments were run.
When it breaks
It is the finding that is uncomfortable, not the method. Across 389 tests, increasing advertising budgets relative to competitors did not increase sales in general, and standard recall and persuasion copy-test measures showed no strong relationship to sales effect — the two things most advertisers optimise failed to predict the outcome. A later analysis of 241 tests run between 1989 and 2003 found the weight effect significantly above zero for established products, driven by the more recent tests, so the 1995 result is partly a finding about a period. It also only ever covered packaged goods on television.
Case
Lodish and colleagues analysed 389 BehaviorScan split-cable television weight and copy tests on matched household panels. Increasing advertising budgets in relation to competitors did not increase sales in general. What did raise the chance of a positive sales effect was changing brand, copy and media strategy, specifically in categories with many purchase occasions and low in-store merchandising. Standard recall and persuasion copy-test measures showed no strong relationship to sales effectiveness.
Lodish et al. — How T.V. Advertising Works: A Meta-Analysis of 389 Real World Split Cable T.V. Advertising Experiments, JMR (1995) ↗Diagram not yet drawn
Two matched halves of one cable town, one half receiving extra weight; the purchase panels underneath as two near-identical bars, with the difference the test was powered to detect drawn alongside the difference it actually found.
In the wild
Unchecked · not part of the tier
What people have written about this tool in the last twelve months. A machine found these and nobody has read them. Everything above this line was checked by hand.
Loading…