Marketers have long presumed that paid search ads are one of the most effective forms of digital advertising. What better time to get your brand in front of consumers than the moment they’re searching for a product?
But correlation doesn’t prove causation. Runners may be thinner than average because they run. Then again, maybe thin people gravitate toward running.
Related post: CNBC: Advertisers May Be Wasting Money on Retargeting
A similar dynamic is at play with paid search ads. Consumers may click on a paid search ad after they’ve already visited your site. We pulled data from a sampling of our own ecommerce customers about how many visitors from paid search were recent site visitors. Turns out 24% of the sampling had already been to the site.
Marketers should care about this because paid search is typically seen as an acquisition channel versus a remarketing channel. To put it another way, many marketers look at sales from paid search ads as sales that would not have occurred without the advertising.
“Advertisers need to be thinking about search in terms of structuring it and optimizing it for incrementality, just like display advertising.”
However, when eBay stopped advertising on Google AdWords back in 2015, the company saw “no appreciable drop in sales”. A possible explanation is that consumers that were targeted by those search ads were predisposed to purchase anyway; potentially just using the search ads as an easy navigation portal to get back to eBay. Granted, this may have happened with eBay because it’s a well-known brand and consumers don’t need to be reminded to go to the site.
Putting paid search to the test
Not every brand can just drop paid search and see no difference in sales. That’s not the recommended approach. The more important point is that advertisers need to be thinking about search in terms of structuring it and optimizing it for incrementality, just like display advertising.
One way to test the incrementality of search is to conduct a geo-based lift study where you segment users out by designated marketing areas (DMAs). There are 210 DMAs in the U.S., many of them are based around cities. Advertisers will randomly choose some areas and show people ads; other areas will be used as holdouts and the people there will not see ads. Ideally these two markets will share similar characteristics. Testing by region allows advertisers to avoid the contamination that can happen when tests are done at the user-level where people move between browsers and devices and sometimes skew results.
Once the ads are up and running in the treatment DMAs and compared to the holdout DMAs, an advertiser is able to calculate the lift in incremental sales being provided from the search ads. Google ran this style of testing for vacation rental marketplace HomeAway, and it found that display ads were undervalued by 51 percent because of last-click attribution.
One thing’s for certain: Many advertisers still don’t know if paid search ads are having a true impact on incremental sales. But why live on assumptions when proof is a lift test away?