A first-of-its-kind study of audio advertising across three major formats found it performs better than video in grabbing attention and generating brand recall, research that could have a positive impact on a relatively overlooked portion of the media industry.

The study was conducted by attention research firm Lumen and commissioned by Dentsu, with Amazon Ads, Audacy, Cumulus Media, iHeartMedia, Spotify and SXM Media participating. The research was part of the Dentsu Attention Economy project, which celebrates its fifth anniversary this year, and compared against a large database of studies mostly on video ad formats generated by that project.

Among the findings were that 41% of audio ads generated correct brand recall in follow-up surveys, compared to a 38% norm for other advertising, mostly video, in Dentsu’s database. The average impact on brand choice metrics was a positive 10% for audio vs. a 6% norm in Dentsu’s database for other ads. Audio advertising produced average attentive seconds of 10,126 per thousand impressions, more than 50% higher than the norms of 6,501 in Dentu’s database of prior studies.

Video still rules budgets

Of course, video still rules ad budgets. TV and digital media, largely video, accounted for 90% of U.S. measured media outlays last year, according to Vivvix, including paid social data from Pathmatics. Radio accounted for only 3%. A small group of advertisers, led by Procter & Gamble Co. but also including video advertising heavyweight Comcast and pharma marketer Pfizer, did substantially step up radio advertising last year. But CPMs for radio at $5 to $6 are well below those for linear TV ($10-$15) and well below those for streaming video at up to $65.

That could start to change given the attention research.

“What’s really great about this is not only did we bring six competitors together, and big competitors, but we also looked at three different formats,” said Doug Rozen, CEO of Dentsu Media in the Americas. “The fact is audio attention is proving to be really high compared to our norms for other platforms, which is counterintuitive, given that you’re often multitasking when you’re consuming audio. But it actually has that attentive engagement more than other channels.”

Having audio attention data embedded into Dentsu’s planning process and planning tools will enable the agency to “elevate the conversations around audio and also different types within audio, to be able to justify that investment to clients on top of what had existed previously,” said Joanne Leong, global head of planning at Dentsu.

Podcast host ads get more attention

Podcasts measured across Audacy, Cumulus Media, iHeartMedia, Spotify and SXM drove the highest attentive seconds per thousand impressions compared to other digital, social and TV benchmarks in the study. Ads read by podcast hosts performed better than traditional audio ads in podcasts at moving the needle on brand preference.

Radio measured across Audacy, Cumulus and iHeartMedia also drove higher attentive seconds per thousand impressions than other digital, social and TV benchmarks. It was 10 times more efficient at generating brand recall per second than average online video ads measured in Dentsu’s Attention Economy project.

Music streaming measured on Amazon Music across voice with Alexa and on mobile or desktop drove the best brand choice uplift for 30-second ads on Alexa-enabled devices than on desktop and mobile devices.

But, overall, there were not huge differences in attention, recall or brand choice metrics across conventional radio, streaming and podcasts, as all tended to outperform video ads on those measures, said Mike Follett, CEO of Lumen.

How the study worked

Respondents in the study were exposed to listening environments similar to their native audio experiences, such as listening to an hour of radio, podcasts or ad-supported streaming. Later, they took Dentsu’s Attention Economy Survey to gauge ad recall and brand choice uplift compared to a similarly recruited control group. Lumen modeled an attention score equivalent to what it gives on visual attention using factors that included passively collected audio listening data, including whether people fast-forwarded through ads where possible, survey results and the type of ad exposure.

It’s difficult for a firm built around eye tracking to measure audio advertising because, as Follett put it, “the ears have no eyelids.” But measures such as post-exposure brand recall and brand choice surveys can help measure visual and audio advertising on an apples-to-apples basis, he said.

Leong said Dentsu worked with an independent research consultant to look at methodologies for the study, including neuroscience measures such as galvanic skin response. Ultimately, focusing on outcomes such as brand recall and impact on preference made the most sense.

“The attention data for this audio study is inferred from the relationship between audibility and recall,” Follett said. “So there’s a missing step, and we’ve had to sort of fill in the pieces on the basis of what we know about visual attention. We haven’t been able to measure the twitches of hair on the inner ear.”

But, on an apples-to-apples basis, recall for audio and visual media have been measured similarly, Follett said. “Audio media are particularly good at getting people to remember the brand.”

There are implications from the research not only for pure audio advertising but also for the value of “sound on” in video impressions, Follett said.

“The flip side of that,” he said, “is that for those media that aren’t sound on, tough luck. It’s very important, and you ain’t got it.”

Part of that may be how video ads are created for sound-off digital video ads, Follett said. “The creative implication is to literally go back 100 years and try to learn from the silent movie stars.”