In a new paper, researchers from Carnegie Mellon University have found that blind network advertisements can increase click-through rates by up to 20% when targeting only the right keywords.
The research is published in the journal Computers in Human Behavior.
The researchers tested the effectiveness of blind network ads on keywords such as “satellite navigation” and “weather” and found that they were able to increase click rates by nearly 50% for both keywords.
They also found that the ads were able not only to improve click-to-click rates but also conversion rates for advertisers who target keywords that are both in their own network and those of other search engines.
Blind network ads are still in their infancy, but the researchers say their findings may help in the design of future campaigns.
They say that in the future, campaigns may need to be more targeted to reach the right audience.
“The fact that the targeted ads are also targeted to a different target audience raises questions about how these ads are being used, and the value of the targeted advertising in these scenarios,” said co-author and assistant professor of computer science at Carnegie Mellon, Mark Roesch.
“This paper provides a valuable test of how to implement blind network targeting in ad campaigns and suggests new ways of thinking about how to address this issue.”
The researchers found that ads targeting the right keyword in a network were able boost conversion rates by as much as 50% on keywords with keywords that were also included in the network’s search results.
This can be an issue in the case of search queries where there is a lot of noise in the search results, said co in the paper.
For example, the researchers found ads targeting “satellites navigation” could boost conversion rate for both “weather and satellites navigation” for advertisers that targeted those keywords.
When the researchers tested these ads on “weather search queries,” they saw that the ad showed up in searches for “sunny” and for “cloudy” which are not keywords that Google searches show up in search results when searching for those terms.
When targeting “cloud-based weather search queries” with the ads, the ads showed up more frequently for “weather cloud” and the ads for “free cloud weather” as well as “free snow” and ads for both of these keywords.
This suggests that ads that appear in a search for “rain” or “wind” may increase conversion rates because they appear more often in searches when searching on those terms, said Roeschi.
“In the future it may be possible to create campaigns targeting keywords in search queries that are not indexed for keyword terms,” said Riesch.
He added that it would be a good idea to take into account the type of search query a campaign is attempting to reach, and to try to include as much of the search query as possible.
“When searching on ‘weather cloud’ for free cloud weather, for example, it may make sense to target keywords in the same search query that are relevant to the weather in question,” said the study’s lead author, Tasha Sengupta.
“For example, if the keyword ‘free cloud’ appears in search for ‘free weather’ and the keyword appears in a ‘free rain’ search query, it might make sense for a campaign to target both of those keywords in order to increase search traffic.”
The study is a proof of concept, which means that it’s not clear how well this kind of system will work in practice, but it is still an exciting step forward in research.
“As more ad networks attempt to integrate their networks with search results to reach their audiences, this research shows that a better understanding of the mechanisms behind the effectiveness and usability of such ad campaigns will help to design better ad campaigns,” said lead author Senguptas.
“Our study also offers some insight into how to design ad campaigns that maximize the performance of search results on search queries.”
The paper, “Discovery of a network-based ad targeting strategy for targeting keywords with unknown relevance,” was written by Roesck and Sengupton, with support from researchers at Carnegie, University of Chicago, Cornell University, and Harvard University.
The paper is available online at: http://dx.doi.org/10.1016/j.cib.2016.04.019