440. Does Advertising Actually Work? (Part 1: TV)

Episode Summary

Executive Summary: The episode investigates whether advertising actually works, contrasting industry confidence with academic evidence. Through interviews with economists and marketers, it argues that much TV advertising is far less effective than commonly believed, with large studies finding tiny sales lift and often negative ROI. The piece also explains why firms still overspend: poor measurement, endogeneity, and managerial incentives.

Main Topics: The central puzzle: does advertising work? (Priority: 5/5): The episode opens by questioning whether the enormous amount spent on advertising produces real sales gains or just correlated noise. Steve Levitt’s retailer case study (Priority: 5/5): Levitt describes a big-box retailer spending nearly a billion dollars on ads, believing TV was more effective than print, but lacking clean evidence to prove causality. Industry perspective from Keith Weed (Priority: 4/5): The former Unilever CMO argues that advertising is increasingly measurable, useful for broad-reach consumer brands, and still essential for mass-market products. Anna Tuckman’s border-market research design (Priority: 5/5): Tuckman explains how neighboring TV markets can serve as a natural experiment to estimate causal effects of advertising, especially for e-cigarettes. Large-scale benchmark study on TV ads (Priority: 5/5): Tuckman, Shapiro, and Hitch analyze Nielsen data across hundreds of brands and find much lower ad elasticity than prior literature suggested, implying weak average effectiveness. Why firms may over-advertise (Priority: 4/5): The discussion explores principal-agent problems, measurement difficulty, publication bias, and endogeneity as reasons companies may overstate ad returns and spend too much. Implications for digital advertising (Priority: 3/5): The episode closes by previewing that online ads may be different because of better targeting, setting up a follow-up on digital ad effectiveness.

Key Arguments: Advertising is widespread and expensive, but spending alone does not prove effectiveness. Without randomized experiments or strong quasi-experimental methods, ad ROI can be impossible to identify because sales and ad timing are endogenous. A retailer’s newspaper blackout in Pittsburgh showed no sales impact, suggesting some advertising may have near-zero effect. Industry executives argue that advertising is more measurable than in the past and remains critical for broad consumer brands. Tuckman’s e-cigarette study found ads increased e-cigarette sales but reduced cigarette sales, showing that ads can shift demand across products. A broader Nielsen-based study found median TV ad elasticity of about 0.01, far below earlier benchmark estimates of 0.15 to 0.2. The study concluded that many brands likely over-invest in TV ads and could raise profits by cutting spending in an average week. Potential explanations include manager incentives misaligned with firm profit, poor measurement tools, and bias toward publishing positive results.

Data Points: U.S. annual advertising spend: more than $250 billion - Amount spent on advertising in the U.S. each year Global annual advertising spend: more than half a trillion dollars - Worldwide advertising expenditure Retailer ad spend: almost a billion dollars a year - Big-box retailer described by Steve Levitt Pittsburgh blackout duration: 1 month - Accidental period when newspaper inserts were not run in Pittsburgh Unilever product reach: 2.5 billion people daily - People who use a Unilever product every day Countries served by Unilever: about 197 countries - Global footprint described by Keith Weed E-cigarette study period: 2010 to 2015 - Data window used in border-market analysis E-cigarette markets studied: more than 200 border markets - Geographic scope of the e-cigarette study Missing cigarette packs without e-cigarette ads: approximately 130 million packs per year - Counterfactual estimate from Tuckman’s e-cigarette paper Historical benchmark ad elasticity: 0.15 to 0.2 - Earlier meta-analysis benchmark for TV advertising effectiveness Estimated ad elasticity in Tuckman/Shapiro/Hitch work: 0.01 - Median brand estimate from Nielsen-based study Interpretable sales effect of doubling ad spend: about 1% increase in sales - Implied by 0.01 ad elasticity Brands in Nielsen sales data: more than 300,000 - Total brand-level sales observations available Top brands initially considered: top 500 brands by dollar sales - Subset used for analysis Brands retained for ad-sales merge: 288 brands - Final sample with sufficient TV advertising variation Observed ROI conclusion: negative ROI in an average week - Most brands in the sample earned higher profits by not advertising that week

Pivotal Quotes: "with the data you have, with nothing like a randomized experiment, it's just possible that the return on investment could be anywhere from zero to infinity" — Steve Levitt: Explaining why observational data from the retailer could not identify causal ad effects "we find that almost all brands seem to be over-advertising and that they are earning a negative ROI from advertising in an average week" — Anna Tuckman: Summarizing the main conclusion of the large Nielsen-based TV advertising study "half the money I spend on advertising is wasted. The trouble is, I don't know which half" — Attributed to John Wanamaker: Used to frame the long-standing uncertainty about ad effectiveness

Implications: For firms, ad spending should be tested more rigorously and trimmed where returns are weak. For consumers, many brand costs may be inflated by inefficient marketing. The episode suggests the advertising industry may be overestimating TV’s average impact, though digital ads may be a different case.