Display advertising in part pays for the internet. The much unloved ad units help fund the content consumers are actually interested in. Banner ads may be disruptive to consumers' experience of the web - but they are not going away anytime soon, so they need to be good, really good. But what makes a good display ad is moving from being an art to a science.

As an experiment Automated Creative decided to explore various aspects of brands' recent online display ads to identify common themes, as well as potential opportunities and threats. Here’s an overview of what we found out:

The most common emotion is ‘joy’
Headwear is cool.
Purple is the new black
Soft selling is the way forward

Let’s have a look at what we found out in detail:

1. The most common emotion is ‘joy’

32% of the ads featured people and 42% of all of the ads featuring people were classified as using the expression of ‘joy’. This makes sense. Why would a brand use an ad of someone looking grumpy? But the question needs to be asked - if everyone is using ‘joy’ then it makes all ads look the same.

Key takeaway - "One of advertising's key tasks is to build differentiation. Look for opportunities to stand out – not to fit in. When briefing your ad design think about other routes than consumers enjoying your product that land your brand’s strategy. It’s worth thinking more laterally about the lifestyle benefit your product brings.

2. Headwear is cool

This surprised us. What is it about banner ads that means headwear is necessary? Do brands feel that headwear helps the consumer understand the context of the image?

Key takeaway - Don’t let your use of headwear in your ads obscure more powerful emotions. After all we’re drawn more to faces than we are to flat caps.

3. Purple is the new black

Of the ads we analysed the most commonly used colour was the colour purple at 12% of all ads.

Key takeaway - Go broad in your colour palette - green may be your brand colour of choice - but no one cares when they are ignoring your ads. Make sure you aren’t just using one colour or doing blunt A/B tests. Systematically testing visual triggers (such as colour) at scale is the only way to understand why your audience click on your ads. Stop guessing and start testing.

4. Soft Selling is the way forward

The phrase ‘learn more’ was used in 17% of all ads but brands still couldn’t resist using ‘Shop now’ 14% of the time. Online advertising is designed to interrupt our experience of the web and it pays for many of the services consumers love and rely on. But it’s rarely appreciated, if ever. So it’s not surprising that brands aren’t using direct sales language.

Key takeaway - There’s an opportunity for brands to go for the kill and be direct in their language. If you want the audience to do something (e.g click the ad) tell them. But this will ultimately be a guess. The technology is now available to generate and test copy at a huge speed and scale allowing brands to get a granular understanding of what written triggers drive performance.

How did this experiment work?

We used machine learning based visual recognition algorithms to find the unconscious themes/biases that brands use in their display advertising. Our AI analysed each ad and broke down what was in each ad in terms of image and copy. We then collated this data and using our Creator tool to understand what the themes were. It’s worth remembering that no matter how powerful AI is - it’s not always right. The images that were recognised using this tool have varying degrees of accuracy. So the data we are sharing here isn’t water-tight but it’s certainly indicative. And it’s also important to understand that we are not reporting on performance (this data isn’t available) but reporting on the emotional triggers that are used most often. This experiment is giving you insight and inspiration into what online advertising looked like recently - as opposed to complete statistical certainty. We use a similar process in our day to day work for clients such as GSK, Diageo, Unilever and BOSE. The fundamental difference between this experiment and our actual product is that we also create the ads at a huge speed and scale. We also optimise their performance by blending human insight and machine creativity sequentially over time. We call this Strategic Content Optimisation.

The correct use of AI is to blend it with human insight. The real magic lies in combining these two skills. Machines are great at doing lots of small jobs quickly and humans are great at doing subjective jobs slowly. Combining human insight with machine creativity is what we do for our clients.

If you’d like to more, please email us hello@automtatedcreative.net

Joy isn’t going to make brands happy