With the buzz around AI-driven tools gaining prominence in the last six months and marketing moving towards the use of various AI tools, Borzo, the global intra-city courier delivery service, released a marketing case study to explore the effectiveness of AI-driven advertising tools compared to those developed by human teams.
In this advertising and marketing experiment, Borzo released two sets of advertising banners on Facebook, Instagram, and Google. The first set was created entirely by 3 AI tools namely ChatGPT, MidJourney and DALL-E and compiled using Viewst, while the second set was developed by the company's marketing team.
Both sets of banners were run parallelly over a month with a total budget of $19,065.
In the development of banners using AI, ChatGPT contributed copy, MidJourney and DALL-E developed creatives and the final output was compiled in Viewst. The marketing team consisted of a team of copywriters and designers.
The paid specialists were common for both campaigns as they optimised and evaluated the results.
The Facebook and Instagram advertising campaign showed that human-designed banners were more effective in bringing in customers, with 9.5x more customers compared to AI-generated banners.
The Customer Acquisition Cost (CAC) of AI banner ads was also 2.5X higher than human-made banner ads. The AI banner ads had a 3X higher clickthrough rate (CTR) compared to human-designed banner ads.
Overall, it was observed that AI-generated ads are more effective in getting attention but fall short on generating clients despite high cash burn. In comparison to this, though human-designed ads draw less attention but are more effective in generating leads whilst burning less.
The Google Ads campaign showed that AI banners with texts had 26% lower CAC compared to human-designed banners with texts and 32% better View-Through Conversion (VTC), despite the fact that AI banners showed a worse CTR.
The campaign showed that human-created campaigns took approximately 3-4 days to be ready for execution taking into account all the planning, previous task load etc. At the same time, AI was more efficient, taking only 4 hours. However, using AI was a tedious task as it required 50-60 prompts to generate the final AD images.
Conversely, human designs underwent very few revisions once the brief was communicated.
“We always look for innovative ways to engage our audience and this time we have taken a brave attempt at testing our marketing team by pitting them against AI,” said Devesh Gangal, Country Marketing Manager, Borzo, India.
He added “Through our recent advertising campaigns on Facebook, Instagram, and Google, we have discovered that while AI-generated banners may have a higher click-through rate, they are often more expensive and generate fewer leads than those created by human designers. It was also observed that the accuracy of output by AI is low compared to Human designers and most creatives generated by AI had vibrant colours instead of brand colours. Though AI generates more attractive and efficient images, human intervention is still required to optimise the final output. At present, AI tools are handy for local entrepreneurs and small businesses, however, bigger brands cannot completely rely on AI but can increase the pace of the entire process. Nonetheless, AI is promising and warrants further exploration. It may take more time for Meta to learn and attract clients with a low CAC for AI-generated ads, but we have not disabled campaigns on Google with AI banners and will continue to work with them. As we continue to refine our approach to online advertising, we are excited to see what the future holds for the use of AI in marketing.”