BUZZVIL BLOG

[Market&Industry] Study shows Rewarded CPM Ads Has Double Effectiveness than Non-rewarded, Hitting ROAS of 1,288 pct

Rewarded ad platform provider Buzzvil currently serves customized ads and contents for 25 million users across 30 countries, including South Korea, Japan, Taiwan, and the United States. Buzzvil’s more than 100 global partners include SK Planet’s OK Cashbag, KT CLiP, Lotte L.Point, CJ ONE, and Hana Members. With premium partners under its belt, Buzzvil allows advertisers to reach a variety of user types.

According to the 2017 report by the Korea Communications Commission, the mobile ad market has continued to grow very rapidly from a zero-percent growth rate in 2010 to 16.7 percent hike in 2016. With the rise of the mobile generation, users are watching the first screen of the mobile- one of Buzzvil’s unique ads inventories- for at least 60 times a day. Therefore, Buzzvil maximizes the ad effect by providing rewards to users who are showing a high level of interest in personalized ads and content. Buzzvil carried out a study in order to check how effective the rewarded CPM ads are in reality, and the results are as follows.

What are Buzzvil’s rewarded ads?

One of the most significant characteristics of the Buzzvil’s ad is that it gives out rewards to users on the lockscreen and various ad inventories. The user can be rewarded just by activating the lockscreen app and consuming personalized ads or contents.

Main Issue: Are rewarded ads effective?

Assumptions and study plans

With regards to the method of attracting users to the online shopping mall by giving out rewards, there are two contradictory assumptions.

  1. The revenue of the shopping mall will increase since more users are accessing the online shopping mall.
  2. The effect of the ad will be minimal since the users will not make a purchase after earning the points.

In order to clarify which assumption is correct, Buzzvil has conducted an A/B test in August and September 2018 with two major commerce businesses in South Korea.

Comparative Analysis of rewarded vs. non-rewarded lockscreen ads

After running an A/B testing with CPM ads, it has been proven that the former assumption is true. We compared two groups of users who were given rewards and those who were not. This testing was carried out on all of our inventories, and it used fixed variables, such as the size of samples and testing period, to obtain statistically meaningful results.


Rewarded ads had more than twice the return on advertising spending compared to non-rewarded ads. Specifically, rewarded ads had 8 times the CTR of non-rewarded ads, but the percentage of actual purchase per click was lower than non-rewarded ads. In spite of that, since rewarded ads have much higher CTR, the percentage of actual purchase per impression for rewarded ads (Group A) were around 1.8-3.3 times higher than the non-rewarded ads (Group B). In other words, as the number of exposure for rewarded ads increased, the results showed a higher level of actual purchase than non-rewarded ones.

Summary

CPM rewarded ads are like a shuttle bus that takes you to the shopping mall for free. To run a shuttle bus costs money, but it attracts more customers. Of course, the number of people who will use the free transportation, but don’t buy anything will also increase.

Still, the beauty of the shuttle buses increases customer inflow to the mall and it can increase the overall revenue. By bringing them there, we can encourage purchase. Similarly, Buzzvil’s ads actively induce clicks by giving out rewards, and as the volume of traffic to the online shopping mall hikes, the revenue increases along with it. The result from the current test is a good example that proves Buzzvil’s assumptions.

Appendix: Analysis of commerce purchase data

Buzzvil offers an effective targeting algorithm to foster high ad efficiency and also gathers insightful information by analyzing the purchase data of users. For the reader’s interest, we will share the purchasing hours and interval based on the data collected in September 2018.


If you look at the purchasing hours in South Korea, the highest distribution occurs from midnight to 1:00 am. and from 10:00 pm to midnight. The interesting fact is that people still make purchases from 3:00 to 5:00 am. It is a good data that shows when we should attempt to communicate with users to trigger purchases. We should know when a certain type of users makes the most purchases.


By looking at the purchasing intervals of users that have made more than 2 purchases in September 2018, 30 percent of re-purchase occurs within an hour after their first purchase. Moreover, approximately 60% of repurchases occur within 24 hours. So it will be effective for advertisers to reach their users right after a purchase is made.