Making Creatives Speak for their performance!
As advertisers began to put a huge chunk of money in reaching online audiences, marketers soon began to adapt to the new ‘Engagement Economy’ where online advertising was not only about reaching users but also engaging them with creatives: be it clicking the creative to land on the website or pushing users through sales funnel. As creative is the only element from marketing mix which interacts with end-users, sharp brains from design to psychology and from content to photography are involved in designing the same.
A creative comprises many elements and like logo, color, text, and image, and the user interacts with each of the elements quite differently. Being in this industry for the last 10 years, and having served millions of impressions for multiple advertisers, we lie at a sweet spot, having a clear look at how users are reacting to each element of these crafted creatives. This blog focuses on understanding the impact of different elements of a creative on its performance, all the possible analysis which can be done on top of that, to help marketers and designers to look at creatives with an eye of data and impact of different attributes in the campaign! Here we have used 180 sample creatives from travel vertical to make a case.
Using open-source Image Processing algorithms, different features can be extracted out of a creative: primarily color, logo and text. And these extracted features can be then combined with each creative’s performance.
Elements of Creative:
We extracted the following features from the dataset considered:
- Logo: Area acquired by the logo on the creative, position of the logo, shape of the logo
- Color: RGB value of top 10 colors present in the creative
- Text: Full written text, the language of the text, the area acquired by the text
Let’s take a look at why these elements are important and how can data extracted can be put to use to generate insights:
1.Color: If you look closely, the most common colors used in a creative for any industry are similar, this is because our brain has certain messaging attached to each color. Choosing an ideal color that supports brand messaging and also has a right impression in the brain of the audience, is part of an aesthetic and part science.
After extracting the most dominant colors present in the creatives considered, the colors were divided into warm, neutral and cool colors based on ‘color theory.’ It was observed that CTR was 4 times higher for ‘neutral colors’ like white and grey, as compared to ‘warm colors’ and ‘cool colors’ like red and blue respectively.
2.Logo: Logo plays an important role in connecting the messaging of the creative to the company. It has a sensory benefit which means they can provide an emotional response that brings consumers to trust the brand. Its mere position in the creative can affect the way it mixes with the messaging and how the audience perceives it.
To identify the position of the logo, each creative can be divided into a 3x3 grid and the position of a logo can be assigned by the same, and the performance of each position can be analyzed.
For the dataset we considered here: no logo was placed at the center of any creative (at position 2,5 or 8). And position 9 saw the highest CTR and was at least 3x the CTR as compared to any other position.
3. Text: Given that the audience has a very less attention span and also they don’t appreciate a creative popping up while browsing online, the text is an element where marketers have to play smart to communicate a strong message and make a user interact with the creative.
After extracting text, the number of characters, number of words, the length of a sentence, can be analyzed concerning the KPI considered. Sentiments of the text and if ‘call to action text’ has any impact on the performance can also be inspected.
In the dataset considered, it was observed that the lesser the number of words, the better was the performance of a creative, supporting the obvious fact that minimalistic messaging intrigues the audience to click and explore.
Additionally, different machine learning models can be applied to classify creatives based on its features and performance, which can help advertisers to identify attributes of best-performing creatives for different verticals!
A tree-based model was built on the features present in the travel dataset and it was observed that the logo-percent area and logo position are the most important features in predicting the CTR of a creative in travel vertical. Logo percent area and logo position combined strongly show a presence of a brand that helps to win the trust (which is so important in the travel industry) of the audience and make them explore what the creative has to offer.
With this strong head start analyzing the basic elements of creatives, marketers can deep dive considering the image aspect. In-depth analysis of colors and text can also be performed looking at which shades and fonts perform better together. With the rise of image detection, data processing, and predictive technologies, there is more than aesthetics in designing creative, and analytics can add a lot of value here. Though designing creatives is more of an art than a science but knowing the reaction of the audience towards different elements of creatives can help teams to make intelligent decisions and run data-driven experiments to make the most engaging creatives.
