Start with what most marketers implicitly believe, because it's the intuition this page exists to complicate: advertising works by convincing people. The classic model is a staircase (attention, interest, desire, action) with the ad as the persuader walking buyers up it. Under this strong theory, the ad's job is conversion of the unconvinced: present the argument or manufacture the desire, and non-buyers become buyers. It's a flattering theory (creative as persuasion machine) and sometimes a true one. But watch what it predicts: attitude change should precede purchase, ad-exposed non-buyers should convert at scale, and stopping the ads should mostly cost you new customers, not current ones.
Andrew Ehrenberg spent decades checking those predictions against actual buying data and found them mostly failing. His alternative (awareness, trial, reinforcement) is the weak theory (weak describing the force per exposure, not the importance): advertising rarely converts the unwilling. It works as publicity: creating and refreshing memory structures, nudging existing propensities, reassuring past buyers, and keeping the brand mentally available for the moment a category need fires. Under this theory, attitude change mostly follows purchase (we like what we buy, having bought it), the heavy lifting happens among people who already buy you sometimes, and the cost of going dark is slow erosion of salience: invisible on this quarter's dashboard, expensive on next year's.
The honest reading of the evidence: the weak theory holds most of it, most of the time, for most established categories: repeat-buying regularities, the rarity of measured attitude conversion, the long-run effectiveness data behind mental availability. But the strong theory keeps real territory. Genuinely new categories and behaviours need teaching (someone had to be persuaded the first fantasy-sports app was worth trying). High-involvement, information-hungry purchases reward actual argument: the central route is real. And direct-response advertising in front of in-market demand (the search ad above the buyer's own query) genuinely converts, because it isn't changing minds, it's meeting one already changed. The mature position is a portfolio: know which mechanism each layer of spend is buying.
Which is why the debate's most useful modern settlement is Binet and Field's long/short split: emotional, broad-reach brand work building memory and fame (the weak theory's machinery) as the base, roughly 60% in their canonical analysis, with targeted, argument-and-offer activation (the strong theory's remnant, aimed at in-market buyers) harvesting on top. The theories stop being rivals and become job descriptions. The catastrophe isn't choosing wrong. It's running one theory's creative on the other theory's budget and judging both by the wrong theory's metrics.