To be honest, I don’t feel I have a definitive answer to this question. If pushed to express my current point of view it would be: 'ads do wear out, to an extent, but not as much or as fast as marketers think and strong ads with plenty more mileage are too often replaced by something less effective.’ As ever, I will revise this POV as I learn more.

Here is a summary of the most relevant evidence I have come across:
  1. If an ad is aired in a concentrated burst, its impact on brand saliency decreases as the frequency of exposure increases (this is referred to as diminishing returns).
  2. For this reason, many advertisers aim to avoid people seeing the same ad more than about 8 times in a month. (This is why having a mix of ads can be helpful).
  3. Ad memories fade over time so an ad’s ability to boost brand saliency comes back if it is sufficiently rested. We know from brand tracking that ad awareness goes up in response to a burst of advertising but falls almost back almost to its base level after just a few (3-4) months.
  4. The work by Dr. Margaret Henderson Blair (see the original paper and this update) indicates that an ad’s ability to increase sales declines as the number of GRPs put behind it during a burst increases: on average, an ad’s sales effectiveness drops to half its original value after about 1200 GRPs. These findings are based on the 'split-cable’ research approach which, I think, does a good job at detecting large, short-term sales uplifts in response to news-driven, persuasive advertising but doesn’t tend to pick up on (i.e. is not sensitive enough to measure?) the slow-building sales-effects of advertising that strengthens the brand’s mental availability (and doesn’t have any exciting product news).
  5. This is consistent with findings (from the ARF and Kantar) that ads delivering 'news-driven persuasion’ decline in persuasiveness as exposure builds - presumably because more and more people have already heard the news. Having said that, since memories from advertising fade, ads delivering 'news’ may still be effective in later bursts by virtue of reminding people about the brand and what it’s good for (i.e. boosting the brand’s mental availability).
  6. There is plenty of evidence that many ads build long-term sales, especially for established brands, simply by enhancing the brand’s mental availability (rather than by persuading people via functional news). Ads working in this way are less likely to wear out. This would explain why the Coke 'Caravan’ Christmas ad has been shown almost every year since 1995 in more than 100 countries. It is still as popular as ever and I suspect the impact on brand and sales impact remains very strong.
  7. People tend to respond to an ad in the same way when they see it repeatedly. They notice and miss the same elements, take out the same messages, feelings and brand associations, and tend to enjoy it to the same degree. (E.g. see this article).
  8. There are, of course, some exceptions. Please refer to 'Marketing Knowledge’ by Dominic Twose for data on how often these occur (rarely) and insights as to why.

Check out this LinkedIn thread for a discussion about the topic.