
Yahoo Native Ads Study
Which kind of native ad can bring the best marketing result?
In September 2015, Yahoo approached NCTU Communication and Cognition Lab hoping we could make recommendations regarding improvements to the ad formats in their news app. They wanted to know the marketing effectiveness difference among static images, non-autoplay video ads, and autoplay video ads.
My Contribution
I led a team of 10 graduate students in psychophysiology track to formulate hypotheses, streamline experiment procedures, conduct surveys, in-lab eye tracking experiment, retrospective and concurrent think-aloud, and in-depth interviews.
Role:
Project leader & usability specialist
Time:
2 months
Team size:
11
Tools:
Eye Tracker
Survey
Interview
Skills:
Experiment design
User testing
Quantitative analysis
Qualitative analysis
Problem Overview
Native ads nowadays play a big part in our daily life, affecting our purchase decision and marketing strategy more than we imagined. In this study, our client, Yahoo, was curious about which kind of native ad can bring them the best marketing effectiveness. We compared 3 different kinds of native ads, including static image ads, non-autoplay video ads, and autoplay video ads. Since NCTU Communication and Cognition Lab is famous for it’s psychophysiological equipments, we combined the advantages from qualitative and quantitative research by using the state-of-the-art eye-tracker, think-aloud, and interview to collect high quality data.

Rationale
In the primary visual cortex, human respond to certain types of information: color, shape, texture, motion, stereoscopic depth. Our visual perception can be separated into 2 stages: the first stage is automatic attentional capture, which can be done without cognition; the second stage is attentional engagement. Since rich media tends to attract people's attention easier than lean media, we proposed that ads in rich media format have a better brand recognition than their counterparts. We used brand recognition and attitude to measure user interaction experiences with native ads.
Research method
A 2-session experiment were conducted. Because of certain limitations, stimuli in the second session cannot be controlled on our end, we used an explorative approach to understand the browsing behavior from users. In the first session, all participants from the experiment group and the control group browsed through 3 different types of native ads and performed concurrent think aloud. In the second session, participants logged in their own Facebook and YouTube accounts to browse through contents. After the trial, they watched their eye-movement videos and performed retrospective think aloud.

Apparatus
An eye tracker consists of cameras, illuminators and algorithms. The illuminators create a pattern of near-infrared light on the eyes, and the cameras take images of the participant’s eyes and the patterns. The image processing algorithms then find specific details in the participant’s eyes and reflections patterns. Based on these details, algorithms calculate the eyes’ position and gaze point on a computer monitor. We used Tobii Pro X2-60. The Pro X2-60 shows exactly where people are looking and, with a sampling rate of 60 Hz, is designed for research into the timing and duration of fixations. The system is perfect for both qualitative research and quantitative research, e.g. calculation of different eye tracking metrics. Participants are easily tracked while moving their head freely without a chinrest in front of the eye tracker, creating a distraction-free test environment.

Procedure
Pre-participation screening
Since Yahoo targeted “mobile heavy users”, we pre-screened and recruited participants via online questionnaires. We measured their mobile usage habit by “The Self-Report Habit Index”, which is the most popular instrument of media habitual usage (Verplanken & Orbell, 2003, p. 1329). Respondents were separated into age groups 18-24 and 25-40 years old. The 12 items include:
- Using smartphone is something I do frequently.
- Using smartphone is something I do automatically.
- Using smartphone is something I do without having to consciously remember.
- Using smartphone is something that makes me feel weird if I do not do it.
- Using smartphone is something I do without thinking.
- Using smartphone is something that would require effort not to do it.
- Using smartphone is something that belongs to my (daily, weekly, monthly) routine.
- Using smartphone is something I start doing before I realize I'm doing it.
- Using smartphone is something I would find hard not to do.
- Using smartphone is something I have no need to think about doing.
- Using smartphone is something that's typically "me."
- Using smartphone is something I have been doing for a long time.
Experiment procedure
When participants arrived in the lab, they were given a copy of informed consent to read again, ask any questions, and sign if they wish to participate. After 5-point eye-tracking calibration, the first trial is to browse through Yahoo news app which was embedded 3 types of stimuli. They had to tap into news article page that interested them, just like they normally would. The second trial was to watch their own eye-tracking screen recordings from first experiment trial and performed retrospective think aloud. After experiment they shared their attitudes and thoughts in the interview.

Findings
Which one would be more interesting, videos or images? While our common sense tells us videos would be more engaging and intriguing, the study result turned to be counterintuitive.
If you want to annoy your uers, try autoplay ads! (No one wants it, right?)
The autoplay video ads attracted most attention among the three types of ad formats, but also most interfering. Autoplay format was rated the least preferred and the lowest on purchase intention. Surprisingly, static image ads were rated most informative and had the highest CTR. Participants also like this format the most.

Yahoo’s non-autoplay video format provides autonomy for the users, while stays as entertaining as possible. However, the cover image of videos were usually not informative enough. According to the interview, some participants were also worried about that videos would use too much of their internet data. We suggested that Yahoo can try gif ads. It combines the strengths of image ads (users can understand the gist at one glance) and video ads (animation is attention grabbing by its nature), but takes less internet data therefor decrease the users’ concern about internet data.
Instead of autoplay videos, gif combines the merits of static image and video formats.
