Helping advertisers understand the value of online video ads

By Alvin Lee

SMU Office of Research & Tech Transfer – According to consumer data company Statista, global video advertising spending will reach some US$37 billion this year, and will top US$45 billion by 2025. Given the impending mass adoption of 5G networks that will make video delivery even more seamless than it already is, those figures might be an under-estimation.

But for the millions of companies and marketing managers paying for video ads, how effective are these ads in generating revenue? Are video ads better than search and display ads? And how do you measure ‘effective’?

Those are some of the questions SMU Assistant Professor of Marketing Li Linyi answers in his research paper “Online Video Advertising: A Dynamic Bayesian Model of Sales Impact, Decay, and Spillover Across Website and Amazon Channels”.

“A lot of companies use per click performance to evaluate online advertising performance,” notes Professor Li of metrics such as Cost per Click, Click-Through Rate, and Conversion Rate. “It is inaccurate of course, but it is much more inaccurate for online video advertising because online video advertising is much more top of the funnel, thus it is harder to track the performance (linking a video ad to a sale) using per click performance metrics.

“I think an important finding of this research is that online video advertising works (at promoting sales), but the effect has a long tail and using per click metrics might result in wrong attribution and thus wrongfully conclude that video ads do not work as well as search ads or even display ads.”

AdStock and the long tail of video ads

Together with Assistant Professor Shyam Gopinath at Indiana University and Professor Steve Carson at the University of Utah, Professor Li examined online advertising data from an American maker of personal care products. Specifically, the researchers looked at its YouTube ads where viewers had the option to skip the ad after five seconds, which is recorded as an impression if it is skipped. If the video ad is not skipped in the first 30 seconds or at all, it is recorded as a video view.

Since it is impossible to know whether the viewer is even paying attention during the 30 seconds, should advertisers insist on paying only for clicks but not views? Or should researchers only look at non-skippable ads?

“The non-skippable format is much more like traditional TV ads,” Professor Li observes. “With the skippable format, you will pay for both views and clicks; you cannot just pay for views or for clicks. Since the vast majority of video ads are using this skippable format (to improve customer satisfaction), I think studying the skippable format is more important and interesting.”

The long tail that Professor Li speaks of pertains to the theory of AdStock that British statistician Simon Broadbent coined in 1979. Often described as “the prolonged or lagged effect of advertising on consumer purchase behaviour”, the theory asserts that advertising has a half-life period by the end of which awareness of advertising falls, or decays, to half of the present level.

By running the sales data of the personal care products company through their model, the researchers found that:

• Video views (of 30 seconds or longer) contribute more to the buildup of AdStock than video or display impressions;
• Even though search impressions contribute more to AdStock than video views, the effects of video decay at a slower rate than either search or display, resulting in a greater long-term sales impact.

The study also found that video clicks are not particularly effective in driving sales, and therefore perform more of a top-of-funnel or awareness/brand-building function.

Implications for advertisers

Professor Li also writes that “video ads increase the effectiveness of other digital advertising channels by increasing the click-through rate for both search and display impressions”, which has implications on the cost and sales attribution.

“One key difference between online video advertising and many other types of online advertising is the vast majority of the time, advertisers are paying money when no customer is brought to the official website. Understandably, advertisers feel a bit insecure and are reluctant to allocate more resources to online video advertising.

“I hope this research sheds some light on the effectiveness of online video advertising and gives managers support for using online video advertising more often. Online video advertising is worth the money you are paying, and the return on investment (ROI) might be even higher than search or display ads; you just need to wait a bit longer to see the results.”

Back to Research@SMU Feb 2021 Issue