fbpx
AdvertisingDigitalEssaysFeaturedInsightsOpinion

Digital Essays 2022: Creating powerful images with data-driven insights – by Ad-Lib Digital’s Janira Hernandez

Can data help advertisers choose images and ensure high visibility for creative ads? Ad-Lib Digital’s Janira Hernandez answers

By Janira Hernandez, head of MEA, Ad-Lib Digital

Marketers are constantly trying new solutions to produce engaging ads that increase ad performance in addition to driving brand awareness. Between product, promotions and people, advertisers have many options while choosing images for creative ads. But how do we know which one has the most significant impact and why?

Ad-Lib.io, a Smartly.io company, has worked with multiple global brands across industries. We decided to analyse creative data to observe how people – particularly faces – can affect ad performance and look into factors that drive consumers’ interest. Our data analysis show that ads with faces often outperform ads without faces. We assessed creative findings and dug into clients’ questions about levers of creative effectiveness, and how we apply analytics to test hypotheses on the drivers of performance. For example, what is the impact of a person’s face on the ad viewer? Do consumers respond with empathy, self-recognition or admiration? 

We observed a 30 per cent increase in click-through rates (CTR) for creative ads with human faces when compare with ads that comprised only copy and ads with generic images. However, adding a face into a creative ad thoughtlessly will not always work. Entropy makes a huge difference here.

What is entropy 

Entropy is a statistical measure of information transmitted by an image. An image with lower entropy typically has fewer colours and less variation. Higher entropy means the opposite – lots of colours and elements come together in a single image to convey a lot of visual information. We analysed creative ads with faces and different levels of entropy to measure their performance impact. The results showed that adding a face leads to higher predicted CTR in simple ads with low entropy. However, adding a face to an already busy ad with high entropy only increases visual clutter and will decrease the ad performance simply because the face on the ad cannot stand out and capture viewers’ attention.

The face-impact

Why does adding a human face to a creative ad make a difference? Faces, in short, are very memorable. We took face-related analysis a step further by applying open source memorability models to advertisements – with and without faces. The memorability models create a heat map with locations where the ad performs well.

On every image, some regions of the image are likely to be forgotten, imagined or hallucinated. The model uses a neural network to consider various clusters of gradients, textures and colour features to highlight regions that are more memorable than others. When analysed and measured with this model, we observed that ads with high visibility and memorability scores incorporate high concentration on the face. Our analysis also shows that fewer, larger faces make consistently more memorable ads than those without any human faces or that display a crowd of faces. When the ad is less busy, the memorability improves significantly.

Memorability is just one of the many creative and media factors that affects ad performance. It is essential to consider recent advances in data and modeling to measure visual impact, especially during pre-campaign analysis. Working with advanced technology such as heat maps can help identify elements that draw consumers’ attention in a creative ad. Too much red on the heat map indicates that the ad may be overly busy, and too little red indicates a lack of noticeable elements in the ad. During a pre-campaign evaluation of creative ads, placing faces, entropy and memorability at the top of your checklist can ensure improved ad performance.

It helps to work with the right technology partners who can provide tools such as heat map analysis. It allows brands to understand and assess creative data before and after campaigns.