Creative & Multi Variable Analysis – AppCampaign - Case Study
In working with a prominent shopping app, Variable Media leaned heavily on data analysis and YouTube expertise to test a direct response marketing strategy with a cost per install as the primary goal. Our client partner was interested in the YouTube audience, and the value users could yield when compared to other media channels as it pertains to lifetime value.
Additionally, with our primary objective set at installs, we wanted to evaluate multiple variables similar to the example on your right comparing mobile platform CPI results.
During the campaign flight and within the case study below you will find two key optimization areas within creative and targeting that help propel performance while providing critical brand insight and platform understanding.
During the campaign, we faced a multitude of challenges. The largest difficulty became evident before the campaign with a strict budget and flight dates. This clamp prevents us from maximizing budget against the most relevant audiences limiting learning and optimization opportunities. With limitations on reaching deeper levels of analysis, we focused on two distinct areas?
Creative – We developed three similar with slightly differentiated long-form versions combined with two different endings. One ending to our video was standard, and one ending provided more information on how to download and implemented the app.
- 5 different bumpers (five-second ad) also were developed and tested from a retargeting, broad, and audience targeted perspective. As we transitioned to the best performing creative in week 3, two different edits were also developed to analyze results.
Targeting – Reaching the right audience would be substantial and therefore a large emphasis would be put on conversion rate and CPA. With 30 campaigns over seven weeks, we were able to implement multiple layers of audience analysis specifically across key variables:
- mobile platform (2)
- targeting type (5)
- target audience (11).
The solution for this campaign would be a smooth spend cadence, combined with enhanced bid optimization in parallel with a creative rotation that leads to the promotion of best performing placements and targets. After a few weeks of media, bumper remarketing would be implemented along with first level variable analysis to determine any targeting optimizations we can make to improve performance.
Six creatives with three slightly different versions and two seperate endings produced a unique performance-driven marketing story in the first half of the campaign. As we optimized to best performing our bumper strategy was implemented and proved useful, educational, and possibly could have been a scalable solution with an extension of the campaign.
The impact of audience and day trading user behavior on performance also proved crucial to campaign performance. With the marketing constraints in the challenge section, our launch began with a very broad target to limit our cost risk while maximizing our exposure. As we narrowed our audience, we did see increases cost per reach. However, we were betting on a conversion rate increase due to a more targeted audience which ultimately led to a lower CPI
By bid optimizing throughout the campaign, we maximized our cost and delivery. Coupled with swift optimization based on learning from creative performance and demographic audience segmentation we were able to improve results drastically. The budget was the only constraint for this campaign as more campaign intelligence and improved performance were both projected if the opportunity presented itself.
- Vid 3- 46% lower CPI
- Vid 3– CTI 2x+ higher that of other two creative
- Vid 3– worst performing VR & CTR, What’s your Primary KPI?
- Non-Parents CPI 47% more efficient
- Non-Parents CTI 3.4x higher than Parents
- 30% of Non-Parents that watched the video 3 ad downloaded the app.
- Average 20% Click To Install Ratio after creative & parent targeting changes
- < $10 CPI by the third week of optimization, below client goal.
- 18-34 Contributed the best performance primarily from the Non-Parent sub-segment