The Campaign

To customize content messaging, the team used a combination of social media insights, Google Vision API, website analytics, and media performance to highlight the popular online conversations. We would pull conversations from relevant hashtags and keywords and run them through the Google Vision API. With machine learning, the app could extract insights from the image, like color, emotion, location. By layering this data on top of real-time conversations and media performance, we could understand what type of messaging we should be using for different audience segments and brands. First we looked at Ramadan-specific insights. These insights were divided into four major themes. After which, we looked at Beauty Insights with five subcategories, and finally Food and Home insights. This process allowed us to understand people had questions like how to break their fast or specific dishes people were looking forward to eating. Our media instantly became more relevant.

Creative Execution

Over the course of two days, 72 hours prior to the peak ecommerce shopping period, we gathered in the media command center (MCC) to produce an ambitious amount of Vogon bumpers. This two-day agile workshop required the help of a core team with 17 people. Throughout the two days, an additional 20 people came through the MCC, providing expertise when needed. We analyzed more than 46,000 images and 276,000 possible data points using the Vision API tool. A photo with just a hashtag is irrelevant to normal social listening tools. But with Vision API, we could analyze sentiment based on the true emotion on someone’s face. This Vision API data was complemented with the learnings with social conversations and paid media performance. The team was able to turn these real-time insights immediately into new creatives. The output was highly contextual, dynamic creative Vogon bumpers.

The campaign accomplished what it set out to do, help Unilever stand out during a crowded Ramadhan season. We created over 300 different variations of creatives, with more than 200 of them being pushed live within 24 hours. Across the 17 brands that participated in the creative workshop, their combined total views was more than 196 million views. Not all views are equal, as we know many ads are viewed in the background. While 196 million was an excellent number of views, were were keen on understanding the interest in the brand and the ad recall the content created. The most important number was the +3.2% increase in ad recall and the 1.6% bump in favorability. Overall, the creative theme behind the entire 1001 Inspirasi Ramadhan campaign saw a 6,646% jump in creative interest.

The focus of this campaign was to use data to make our content and media more relevant. We didn’t want to use high level data or annual trends, but rather actual conversations that were happening in real-time. But combining the efforts of media strategists, data analysts, and creative copywriters, we were able to use data to separate our media from competitors using more generic insights to drive more awareness and action with our paid media.

Insights, Strategy and the Idea

Our strategy was to use these real-time insights to create dynamic creatives and improve our audience targeting. We chose YouTube as the core media channel and six-second bumpers as the best format to quickly deliver contextual messaging. This format would allow us to be agile with content creation and making the dynamic creative even more relevant. The teams picked up general themes and coupled these with specific brand insights to develop the new, engaging variations of the Vogon bumpers. We then used these same insights to create different audience buckets and keywords to target, per audience. For example, our corporate Ramadan campaign had 20 different audience segments and approximately ten targeted keywords per segment. This created very relevant copy for everyone we delivered media to.