How google’s media lab boosts youtube ad results

In a world in which most Americans are exposed to up to 10,000 ads each day, its clear that one ad impression isnt going to cut through the noise. Yet brands also have to be wary of overexposure. When people see the same ad over and over, it can lead to frustration and decreased brand equity. This challenge is one we at the Google Media Lab, which manages the media strategy for all of Googles marketing, have spent a lot of time thinking about, experimenting on and workshopping.

When it comes to optimal frequency and great storytelling, weve found an unexpected solution in video ad sequencing on YouTube. In addition to improving consideration and purchase intent, were seeing impressive results when it comes to delivering frequency at the right level on a per viewer basis. In dozens of campaigns across the U.S., U.K., Germany, Korea and Japan for Google brands, including Stadia and Chromebook, video ad sequencing has driven 80% to 110% higher brand lift and 50% to 70% lower cost-per-lifted-user than non-video ad sequencing campaigns, depending on the marketing objective.1

Heres what weve learned along the way:

Finding the frequency sweet spot

As TV audiences shift to streaming, hitting target rating-point goals on TV means driving frequency way up, to the point of diminishing returns. As consumers seamlessly shift between devices and channels, this approach can mean they are over- or underexposed, depending on their viewing habits. Some consumers receive too many ads, which can lead to annoyance. Others receive so few that theres minimal impact. Much of this is imperceptible to media planners, since tools and targeting often rely on average frequencies.

When it comes to frequency, theres no magic number; brand awareness, product maturity and campaign goals all play a factor. Most frequency solutions limit exposure via capping tools without the option to control or increase frequency when needed at the viewer level. With video ad sequencing, campaigns can be uniquely optimized to drive higher frequencies for maximum impact, specifically with your target audience on YouTube. We found that running simple creative and audience experiments is an affordable way to find the frequency sweet spot for different markets and brands.

Across all our Google campaigns from April 2020 to December 2020 that used video ad sequencing, higher frequency drove higher absolute brand lift.2 On average, ads shown in frequencies of three or more drove 61% higher absolute brand lift than frequencies of one and two.3

Credit: YouTube

Expanding storytelling potential: variety plus frequency

Were all wired to remember great stories, and video ad sequencing allows us to tell stories to customers across their purchase journey. At Media Lab, we love experimenting with the storytelling potential of sequencing; its proved to be more memorable than the 30-second ad spot.4 In testing five story sequences, we found that creating a narrative structure that built on itself — responding to what people watched or engaged with previously — performed best, driving increases in brand awareness, ad recall and purchase intent.

We also discovered that three increasingly advanced sequencing methods drove higher brand lift, respectively.5

Credit: YouTube

Three methods for using video ad sequencing to drive higher lift

Scatter plot shows that brand lift grows as complexity of video-ad sequencing methods increase. Creative orchestration and storytelling are the recommended methods, with middle and high complexity and correlating lift. Single-asset scored lowest in both.

We recommend using the creative orchestration and storytelling methods for maximum impact. For our 2020 Brand Helpfulness” program, we ran two video ad sequence campaigns. One showed the same 60-second hero film —Thank you, teachers. Thank you times infinity” — three times. The other used three different creative variations: the hero film, followed by the shorter stories: “Teachers Appreciation Week: ABCs” and Teachers Appreciation Week: Syllables.”

The second sequence, the one with more creative variation, not only drove higher ad recall but also a 60% higher completion rate, 9% lower CPM, a 16% higher view rate and a weekly and cumulative frequency closer to our goal than the single asset. We continue to see these same video results play out across different brands and markets. Variety wins.

Adopting a test-and-learn approach to measurement

At Media Lab, we pride ourselves on being digital-first marketers who want to push the industry forward. Having a test-and-learn mindset is a belief we hold so dear that we reserve 20% to 30% of every campaign budget for testing.

YouTube is one of the most data-rich and data-quick platforms. Our brand-lift surveys get thousands of respondents, allowing us to slice and dice lots of data in many different ways. And because the surveys are completed quickly, we are provided with fast, efficient, statistically significant results. This means we can test a lot of theories and ideas cheaply on YouTube, and then scale those learnings to ensure maximum impact and good budget stewardship. We are constantly tinkering and fine-tuning throughout the duration of a campaign, which helps us optimize video results.

Experimentation is embedded into the DNA of Media Lab, hence the name. Were excited to share our findings on sequencing, and to continually adjust, adapt and learn, in everything from the tools we use to the way we watch evolves.



1Google internal data, based on Google marketing campaigns measured by BLS2.0, 2019–2020
2Google, Global, VAS meta analysis data from 14,615 campaigns and 554 advertisers, April 2020–Dec. 2020
3Google, Global, VAS meta analysis, data from 14,615 campaigns and 554 advertisers, April 2020–Dec. 2020
4Google/Ipsos lab experiment, U.S., n=7,500 people ages 18–64, Nov. 2018
5Google, Global, VAS meta analysis, data from 14,615 campaigns and 554 advertisers, April 2020–Dec. 2020