The survey, whose findings are part of MarketingSherpa's 2008 Online Advertising Handbook, polled a variety of marketer clients and agencies about their online advertising tactics, what they found worked and what they found didn't work as well.
When asked which tactics most improved ROI, 25% of respondents said online ad effectiveness studies increased their online ad returns significantly. That was followed by online focus groups at 23%, upgraded site analytics software at 21%, eye-tracking studies at 20% and multivariate ad testing at 19%.
"It comes down to good advertising is good advertising," said Tim McAtee, a MarketingSherpa analyst who wrote the report. "Good advertising works at a psychological level. When we asked people what tests most drastically affect ROI, it was qualitative tests that affected that."
Marketers, he urged, should focus on ROI rather than clicks, be more aware of how creative elements encourage actions, and match up design tactics with those objectives. For example, if a marketer has a complicated message, video should be used to convey it. Or if the goal is to have people sign up for an e-mail providing more information, then an in-banner registration box should be provided.
The report also polled advertisers on the kinds of targeting and placement tactics that were working best and found that contextual targeting was the most effective, with 41% reporting that it delivered great ROI, followed by behavioral, with 37%. Almost 30% of respondents reported great ROI from text-link ads and another 23% endorsed affiliate marketing.
"When you mix the quality with the best of targeting -- behavioral or context -- that's when you really see quite an extraordinary potential for the medium," said Stefan Tornquist, research director at MarketingSherpa.
Still, one of the barriers to marketers using more qualitative measures is figuring out how to marry them to the quant-based measures.
"We're not seeing marketers mix the two and create dashboards that show both the qualitative and quantitative side of things," said Mr. Tornquist. "Part of that is people who have been trained to do these tend not to be trained in both sides of it. They're either the crunchy granola ethnographers or math-based analytics data cruncher."