When online display ads work too well

Display ads can prompt consumers to search for other brands, research finds

By Robin Mordfin     
July 01, 2013

From: Magazine

Companies run online display advertisements to increase sales. But it turns out that in some cases these ads may be increasing sales for competitors.

Dan Nguyen, a doctoral candidate at Chicago Booth, and Randall Lewis of Google measure how display advertising affects searches for brand-name products by looking at The duo recorded data on three days, during which time advertising was split between two advertisers. One ad was shown to visitors who arrived on an even second (as in, at 3 p.m. exactly), while the other ad was shown to those who arrived on an odd second (one second after 3 p.m.), so that every user had a 50 percent chance of exposure to either ad. On each day, one display advertisement advertised a product from a specific category, such as mobile devices, while the other advertised an unrelated product.

The researchers tracked activity on the Yahoo! search engine, including what visitors searched for and what they clicked on, for up to 10 minutes after the visitors saw the display ad. That includes the number of times people clicked on links from the search result page, as well as searches they performed on the website, with the researchers noting whether the visitors searched for the brand advertised, related brands, or the product category more generally.

Nguyen and Lewis find that display ads caused an average 30-45 percent increase in searches for the advertised brand and an average 1-6 percent increase in searches for competitive brands. They also find the ads caused an up to 23 percent bump in searches for some competitors' brands. They conclude that display advertising can raise the likelihood that consumers will search for many brands in a product category.  

Of the campaigns studied, the display ad for the Samsung Galaxy Tab, a tablet computer, helped competitors the most. If the Galaxy Tab campaign had run for one full day, it would have produced 424 more searches for the brand than would have existed without the advertisement, the researchers believe. However, the display ad would have provided an even bigger boost to the Galaxy Tab's most popular competitor, Apple's iPad, as Lewis and Nguyen estimate it would have produced an additional 857 searches for the iPad. The Galaxy ad, they conclude, was more effective for the iPad than it was for the Galaxy.  

A display ad for insurance company Progressive, however, did not noticeably increase searches for its competitors. The researchers surmise this is probably because insurers' websites quote competitors' prices. Therefore, if a person saw the Progressive ad and subsequently used Progressive's website, he would see prices for other insurers, thus obviating the need to search competitors' sites for additional information.
Lewis and Nguyen's research is the first to use a natural experiment to identify the positive spillovers of brand advertising on the whole product category. Advertisers may wish to adjust their bids in search ad auctions—or their search ad strategy—accordingly.

Randall Lewis and Dan Nguyen, "A Samsung Ad and the iPad: Display Advertising's Competitive Spillovers to Online Search," Working paper, April 2013.