The study, conducted for the AOP by GfK NOP, used an online survey of 1,340 regular internet users in the UK to look into ways of measuring engagement with online content and ads. It segmented websites into those carrying original content, including newspaper and magazine sites, portals such as Yahoo and MSN and social networks.

Adverts featured on the original content sites scored highest for trust and for influencing positive brand opinion, while social networks scored lowest on both counts. “Trust is an extremely important metric for driving advertising response and is yet to be strongly associated with social networks,” the study says.

Lee Baker, director of AOP, said: “This is the first research to highlight the inadequacies of metrics to measure engagement that the industry currently employs.”

Tim Cain, head of research and insight at the AOP, said the research “sought to understand better the underlying factors that can really ascribe engagement and create a framework for determining how and where it exists… This study has created a model that can be replicated for comparative analysis at individual site, category or genre level.”

The AOP’s white paper says that volume alone cannot be used to measure engagement. “It is the quality of the time spent on the site, and perceptions of the experience that contribute to overall engagement and receptivity to the advertiisng contained on that site,” it says.

Sarah Messer, chair of the AOP’s research committee and head of commercial research at ITV, said: “This research shows that some indicators, such as ‘dwell time’, are not an accurate measure of ‘engagement time’. Users will visit social network sites daily and for long periods of time, yet by visiting a b2b, consumer or business site and obtaining relevant information or news quickly, doesn’t make the user any less ‘engaged’. In fact, we’ve found users are more likely to respond to advertising on content sites than social networks.”

The study mirrored work carried out by the Online Publishers Association in the US, which used a similar approach of asking respondents to rate sites on a range of attributes, as well as how they would react to ads on those sites.