AI and the Future of Ad Creative Automation
This is a joint post written by Amir Harel from Business Solutions Engineering and Karen Chui from Partner Management
We know that creative is important for ads.
There’s a century old creative ad industry and countless award shows every year for it. It’s basically the best part of watching the Superbowl.
For decades, marketing campaigns were are exclusive for larger brands, because creating a campaign took months of planning and an army of experts. Creative production was also not something that all advertisers can afford.
Meta has democratised advertising to all businesses, so every business from individual contractor to large enterprise has the opportunity to reach the audience they need. But what about the other half of advertising, the creative?
How can we democratize creativity like we did with media?
Creative AI is the technology that will expand access to creativity for every individual and businesses. The future of creativity is powered by an AI assistant who can assist you to write creative briefs, to automate A/B testing, creative production and creative optimization.
Automation isn’t a threat to creativity; it’s a tool that allows more people to create. Just like a calculator, it didn’t kill math, did it?
With the advancement of technology, artificial intelligence (AI) enables automation in the creative process; making the ad creative journey more efficient and available for more types of advertisers. It is becoming a strong force in the computing world to solve and to predict all kinds of complex problems that used to be subjective and challenging to do before.
In this blog post, we would like to share how creative AI and automation can remove current roadblocks for small businesses.
Let’s start by looking at the creative process:
The above diagram illustrates a basic creative process of all the decisions an advertiser has to make in order to create a successful campaign. These may be quite daunting for small businesses who are creating ads the first few times.
Here’s where Creative AI can help. A few of the steps above can already be automated today; and we can see many more being available for automation in the near future!
What can we automate today in creative production?
There are a few platforms that can automate part of creative production. They are not perfect (yet) and they all exist as separate products at the moment. But in-time, we see them coming together for marketing purposes:
- Generating Copy (headlines/Call-to-action)
Copy is the text messages that appear in an ad. Big companies invest a lot of time and money in finding the best copy test for their ads because they have big impact in performance. However, this is not something small businesses can afford. This is where AI comes in, by analysing copy text, businesses, and ads performance, AI can automatically generate copy text for ads.
In a recent Hackathon we had in Meta, a group of Business Solutions Engineers created a project exactly focused on this problem. They trained a language model using +100K ads; by extracting keywords from images, using ad metadata and ad copy they built a model that can predict the next words in a text or generate the whole text for an ad copy. This is a similar technology as seen on the autocomplete feature on other platform’s mail but focused on the ads copy text.
As the technology will mature, I think we will see more solutions in the future that solve the copy text for businesses in ads.
2. Generating visuals from product catalogue
Ecommerce businesses usually have a product catalogue holding plain product visuals and tags. Plain product visuals are not ads. You need copy, logo and call-to-action. With creative AI, it will suggest video or image templates where they can turn your product catalogue into ads in all formats. The more advanced ones can even render background images and videos that can be generated by changing variables, like different background for Christmas, black Friday sales and valentines day.
3. Generating the visual (Images/videos)
A picture is worth a thousand words. The image of an ad catches the eye and can quickly trigger emotion and sentiment to lead to action. Brands are willing to spend millions on producing the visual storytelling in an ad. So how can we help the businesses with a smaller budget?
This is where AI can come into the picture (no pun intended). Generating images from thin air can solve these problems. There are already examples of such technology — thispersondoesnotexist.com — is a website that generates a person face image which doesn’t exist. The algorithm behind it is trained on a huge dataset of real images, then uses a type of neural network known as a generative adversarial network (or GAN) to fabricate new examples.
As this technology will evolve, we can expect to see media assets that are generated automatically based on some input. This can be a huge game changer for small businesses who don’t have the resources to invest in production and editing.
4. Generating creative variants for testing
As creative AI helps small businesses to produce creative quickly and automatically, it also gives these businesses the time and ability to test to see which one of these work best for their businesses. Instead of playing the guessing game, they can generate 4, 6, even 10 creatives and to split test them to see which ones performs the best.
In order to enable more automation in creative production, it’s pivotal that we use creative AI to learn the creative insight of what’s worked and what didn’t work; so they can make better recommendations and decisions in the future.
Creative insight is the foundation of automation ad creative.
Creative insight capabilities today:
Successful creative used to be considered as art, and not science. One of the first thing that comes to my mind when I thought about AI and creatives, is to solve the question that advertisers have. How will this ad creative perform? How can I make it better?
The more advanced creative AI solutions can predict how well an ad will perform. The AI trains a model with a given creative and target audience, what would be the likelihood of this creative to drive high performance for the ad. This technology will allow advertisers to save money by running A/B testing and be able to optimise creatives by testing thousands of options to find the best creative for a certain ad.
Creative insight capabilities today are:
- Creative audit for ad account (can be single ad account or across multiple ad account)
- Brand governance with custom tagging
- Creative performance prediction with machine learning
- Post campaign analysis to see what’s worked, what didn’t work and why
- Basic creative optimization recommendations upon ad account and/or campaign analysis
While the above capabilities are already adding value to advertisers, they are only in their early stage, and it will become more and more advanced and at some day will probably become a commodity to validate creatives before using them in ad campaigns.
The future of creative automation using AI : End-to-end creative optimization and production
As mentioned above, creative AI’s are predicting campaign performance. The next step to this is automatically generate suggestions on how to optimize the creative for optimal performances; or to combat ad fatigue. So in a way, in the future advance AI system will only require some initial data, in which it will generate campaign creative, monitor and analyse the performance, and will autogenerate improved creatives to push performance up and avoid ad fatigue. All that without a single human intervention. This will allow businesses with limited resources to market their product and services online.
Talk to Us
Whether you are working in AI or Ad creative, or even if you are working on the next big startup of AI Ad creative, we would love to hear your thoughts either here in the comments, or feel free to reach us on LinkedIn: Amir Harel and Karen Chui
