Approaching The Holy Grail of Marketing
Generative AI has the highest likelihood of any technology in our lifetime to unlock marketing’s holy grail - true personalization at scale.
Generative AI will allow marketers to go beyond coarse group-level personalization (e.g. trying different landing pages for men vs. women) - these will be customer journeys personalized for the individual.
Ad images with a high likelihood of conversion will be generated on the fly to resonate with a specific potential customer.
Landing pages and copy will be altered in real-time to meet the customer where they are in their journey.
Customers will be shown the right second product for them at the right time for them with messaging specifically crafted to nurture them.
It is going to give marketers and designers superpowers.
Simply put, the use of generative AI tooling is going to become table stakes for ecommerce brands in 2023 and beyond.
Why is this happening now?
Two opposing forces are colliding: the second order effects of the ATT/iOS14.5 headwind with the tailwind of two technological breakthroughs, large language models (LLMs) and image generation models (also called diffusion models).
LLMs (like ChatGPT) generate text and Diffusion models (like DALL-E or Stable Diffusion) generate images.
Everyone in the industry knows that ATT made it harder to sell online. The policy released as part of the iOS14.5 update limits the amount of data that advertisers and retailers can collect about their users, making it more difficult to deliver targeted ads and personalized product recommendations.
In the wake of the roll-out, CAC has spiked, conversion rates have been hit and measuring the effectiveness of marketing efforts became sorcery instead of science.
The response of the two major ad networks to ATT has been to try to move to a model where the networks themselves try to control as much of the targeting and campaign management as possible. Google’s and Meta’s push towards PMax and Advantage+ respectively is part of this strategy. In order to try and compensate for the conversion data loss caused by ATT, the platforms are trying to automate most of the levers that marketers currently use to optimize acquisition. Meta’s and Google’s reasoning is that they both have more capacity on their side to build complex machine learning models to compensate for iOS14.5 than any individual marketer or brand could.
As we see this trend continue to play out, the highest leverage a brand or marketer will have to drive performance will be in the creative (images) and copy they use - creative and copy will be the new targeting.
This is where generative AI comes in.
Generative AI, a subset of artificial intelligence, is a technology that can create infinite new and original content based on a user prompt and other input data. Depending on the use case and the technology used, this original content can be generated in real-time or on the fly.
Here are three ways that this technology can be used today to improve the customer experience and counteract the iOS14.5 headwinds:
Ad creative image generation
Generative AI can be used to create “lookalike creative” - by analyzing ad creative performance history, 1st party data, and demographics, generative AI algorithms can create unlimited personalized images in seconds that are likely to resonate with specific groups or hit specific goals.
For example, suppose I give a generative AI system the prompt “create 20 different creative concepts for my moisturizer cream that will lead to higher CTR among women 18-25”, the tool can understand what visual elements lead to higher performance according to those parameters across your brand’s ad library (and across the industry) to help you automatically produce more of them.
So, if the system detects that images with a beach setting perform better among women 18-25, you will be able to generate and refine as many of those kinds of images as you like until you find concepts you want to test.
👇 These were made with Treat AI
These images can be: pushed directly into Advantage+ or PMax to run rapid tests as to which creative angles drive uplift, used as part of the creative brief for the design or the video team, or be used to give guidance to influencer or UGC teams as to what to prioritize (e.g. ‘target influencers who can shoot content at the beach’).
Full disclosure: this is where Treat, the company I co-founded, is building out tooling.
Generative AI tools are already being used in e-commerce to create original content such as product descriptions, social media posts, and blog articles. By analyzing existing content and customer data, generative AI algorithms can generate new content that is tailored to the brand's voice and style. This saves time and resources for the brand while ensuring a consistent and high-quality content output.
(btw, all of that paragraph was written by ChatGPT, a product from OpenAI).
What will a brand need to truly leverage the power of Generative AI
1. An enriched understand of who their customers are
As the cost of rapidly creating content (images and copy) drops to close to zero, the value of creating ‘good content’ will increase dramatically. Brands will need as much 1st party, behavioral, and demographic data about their customers as they can get to be able to really personalize the images and copy that people see. Brands should strongly consider partnering with vendors who either have these capabilities pre-built or can help acquire this data to drive the most value from these technologies.
2. A willingness to experiment
These tools are cutting-edge and will still need human supervision. However, most underlying models should get better with usage and produce results that are more specific to a particular brand overtime. Brands looking to get a head start should jump in and start trying things to calibrate the most valuable use cases of this technology.
The Holy Grail
The end state for these tools is that AI will be used to generate personalized high converting shopping experiences for every customer as an individual. Two different customers will see different ads that bring them onto a site, different landing page copy optimized to resonate with them specifically, they will be taken on personalized up-sell and cross-sell journeys, and will be sent different marketing messaging that anticipates their needs and wants. For brands that invest early in testing these products, this should increase customer satisfaction, customer loyalty, and sales.
As brands start experimenting this year, I think we’ll see several finally start to crack the personalization code and pull ahead of the pack. The race is on.