Creative is the #1 performance lever for digital ads. In fact, a study by Google showed that 70% of campaign performance can be attributed to the creative. When this study was first published in 2015, it was used to convince performance marketers to care about ad creative. Today, convincing is no longer needed – marketers regularly use creative insight and optimization tools to maximize their creative ROI.
While AI feels like a new hot topic, a recent Forrester Research survey showed that 68% of marketers are already adopting generative AI capabilities to optimize their creative. At Smartly.io, we’ve also shown that the hype is real through performance-based case studies.
Here’s how we see AI’s potential to radically transform the creative workflow:
- Use data in planning and ideation. Today’s image/video/text understanding models have been trained on billions of data points across the internet. As a result, they’re incredible at understanding what’s going on in an ad, even at a conceptual level. Combine these creative tags with your ad performance data and put it through a recommendation model, and you can use AI to know – not guess – what will drive performance in your creative. Backed by data: across 10 tests with a client, we saw a 45.7% lift and improvement on 9/10 tests when using AI-based creative insights.
- Get to a first draft faster. With generative AI, the gap between idea and ad closes everyday, and doing becomes almost as fast as thinking. Today, you can use AI to reimagine your products in new scenes with AI, generate copy and design ideas, then put it all into an editable template. When working with video, you can do everything from generate scripts and audio, to collage clips together based on theme, to make talking avatars. Backed by data: Meta found that advertisers expect to save five or more hours per week by using generative AI, equivalent to one month per year. (source).
- Scale, localize, and personalize. Once you’ve developed a brand-qualified first draft, AI can help you enhance, localize, and personalize at scale. Plug in a product catalog and generate custom value propositions for each item, adapt to audience interests (as informed by your creative insights), translate audio, scale your design across platforms, and so much more. Backed by data: TUI used Smartly.io’s creative AI to generate tailored reasons to go to all of their flight destinations, which led to a 10% CPA improvement on their video ads, along with hours of time savings
- Optimize and iterate. Once you’ve put the effort into creating a launch-worthy ad, you want it to perform for as long as possible. AI can help you identify signals like fatigue or opportunity for iteration, suggest changes, and even help you execute on them. New variants can be clicks, not hours, away. Backed by data: Meditopia used Smartly.io’s creative optimization tools like ad rotation to cycle ads in and out of time. Using this, they increased ROAS by 20%, and increased conversion rates by 36%.
- Use English as your programming language. In the same way that the gap between idea and ad creation closes with AI, the gap between question and answer does, too. Rather than needing to learn complicated UIs or dig for data, AI will increasingly be able to understand your asks in plain English and query data for you. At Smartly.io, we’re starting to test this in small ways in reporting, creative tagging, and more. Backed by data: credits to the quote, and the supporting LLM research papers, can be found on Andrej Karpathy (Open AI)’s account.
When you put all of these individual AI assists together, you might be able to picture a more radical workflow transformation in the future – advertisers can input their assets and goals, and receive back complete, launch-ready ads. Post-launch, they can watch their ads adapt to post-launch signals, mostly hands-free. That way, you can always make sure that you’re serving the best possible ad for your message and moment in time.
On the path to this AI-driven future, it’s important to keep a few things in mind:
- Review and cleanup is key: There’s a lot of variability in the quality of AI outputs – you’ve probably heard issues about ChatGPT “hallucinating” false information, or image generation tools outputting inappropriate content. At Smartly.io, we’re building governance tools to check outputs automatically, but we believe it’s crucial to keep a human in the loop, especially to check more sensitive things like image generations. To help with this, we offer a managed cleanup service, as well as the ability to review at every stage.
- Incorporate your Brand Voice: text generation models often have a standard voice that sounds a bit “AI”. Luckily, AI can be trained on your brand voice and style and keep those things in mind. Make sure you look for tools that learn from your existing data, to make sure that outputs look and feel like your brand and continue to learn with you.
- Keep humans in the pilot seat: As the CEO of OpenAI, Sam Altman, says: “these [AI] systems are much better at doing tasks than jobs.” (source) A creative mind can now get 10x more done as the gap between idea and execution shrinks. However, AI today is still a “co-pilot” that is only as good as its “pilot.” Work with knowledgeable creative teams to come up with everything from big-picture ideas to the prompts that will help you execute on them.
It’s a very exciting time to be working in advertising. At Smartly.io, we’re lucky to be at the forefront of the industry: we’ve been building in creative scaling, media AI, and more recently creative AI since our inception a decade ago. As we incorporate the recent advances in generative AI, we partner daily with thought leaders like Forrester, as well as our own platform partners (Meta, Google, TikTok, Snap, Pinterest, Twitter). If you’d like to test our creative AI tools or share your perspective on the industry, reach out to your Smartly.io representative or request a demo.