The Advent of Generative Personalization
We’ve been talking a lot about personalization with our good friend and digital strategy extraordinaire Arif Mansuri. Eventually we decided collaborate and document our thinking. Hope you find it useful, and thanks Arif!
Traditional personalization in marketing has been with us for years—we've all received emails with our names in the subject line or seen product recommendations based on our browsing history. But today, we stand at the threshold of something far more transformative: generative personalization, or what might be called "a segment of one."
At its core, generative personalization leverages both customer data and generative AI to create truly unique, dynamically-generated content tailored specifically to each individual.
The Carvana campaign marks a real moment in this evolution. Unlike the recommendation engines or curation engines we've grown accustomed to from services like Spotify Wrapped, Carvana combined customer data with generative AI to create 1.3 million unique, AI-generated videos celebrating individual car purchases. This was personalized content creation at scale, merging video, text, and voice in ways previously unimaginable. We think the implications of this experimental campaign are far-reaching.
From Formulaic to Fantastic Creative
Consider the state of "personalized" marketing today. Most of us have received outbound emails following a predictable formula:
Dear [name],
I see you went to [college name].
Do you need [product/service]?
Or we've seen ads displaying a grid of products we recently browsed. These approaches, while technically "personalized," feel mechanical and often miss the mark on creating meaningful connections. We think AI will enable the industry to do much, much better.
The upside of true generative personalization could be massive. According to McKinsey, personalization at scale often delivers a 1-2% lift in total sales and can cut marketing and sales costs by 10-20%. The impact can be even more dramatic in specific cases—Orangetheory Fitness's hyper-personalized campaign prompted 45,000 class bookings in its launch week, achieved 97% class attendance, and led to the lowest member churn rate in the company's history. These numbers tell a compelling story: tailoring content to the "segment of one" can fundamentally transform customer relationships and business outcomes.
Channel Surfing with Purpose
One of the most promising applications of AI in marketing lies in its ability to accelerate experimentation across emerging channels. Nimble brands who embrace new platforms early have reaped outsized returns in the past:
As Instagram's ad platform grew crowded and expensive, DTC brands like Curology, Brooklinen, and MeUndies flocked to Snapchat's then-new Story Ads
MeUndies shifted one-third of its marketing spend from Facebook to podcast sponsorships and subsequently sold 9 million pairs of underwear from 2015–2019
Many DTC marketers turned to TikTok when big brands were still absent, benefiting from lower CPMs and higher engagement
AI dramatically lowers the barriers to this kind of agile channel experimentation. Generative AI now allows marketers to produce ads and branded content tailored to each platform in a fraction of the time it once took. According to industry surveys, 41% of advertising professionals cite content creation as the area where AI has most improved efficiency, with AI now able to automatically generate unlimited ad variations from a single original asset.
This means marketing teams can quickly produce content for newly trending platforms without the usual budget and lead time constraints. When a channel suddenly shows promise, AI-enabled teams can capitalize immediately by deploying fresh creative while it's still "hot."
Too Much of a Good Thing
There's a fine line between personalization that delights and personalization that disturbs. Brands must be careful to avoid over-investing in capabilities that yield minimal returns. Target's experience is instructive here - their direct response team tested customizing pet food coupon images. Based on the recipient's dog breed, the coupon featured a picture of a big dog or a little dog. Surprisingly, their tests showed there were no discernible impacts on redemption rates between the experiment and control.
More concerning is the risk of consumer backlash. Target's infamous pregnancy targeting incident, where they inadvertently revealed a teenager's pregnancy to her father through targeted coupons, remains a cautionary tale. As brands rush to explore what they can do with AI-powered personalization, many fail to consider what they should do.
Recent consumer research underscores this concern: 75% of consumers find many forms of marketing personalization at least somewhat "creepy," and 22% have switched to different brands following overly intrusive personalized experiences. This phenomenon has led to widespread marketing fatigue, with 70% of consumers unsubscribing from at least three brands in the past quarter due to overwhelming message volume.
Finally, there’s the risk that marketers, in the rush to blindly adopt gen AI, collectively flood consumers with poorly executed, AI-generated slop. The digital experience of ad-supported channels would quickly degrade, and brands would erode the trust of their consumers.
The Future We All Want
There’s a lot of AI slop that’s coming our way, whether in the form of more believable deep fakes, spam and yes, advertising. But we’ve yet to meet a marketer who aspires to contribute to that problem.
Figure 1. AI Slop, as envisaged by Flux 1.1 Pro
Our ideal future of generative personalization would look markedly different from today's landscape. The overall digital experience will be improved for consumers, who will receive relevant and valuable ads and content. This advertising will feel natural and native to each platform, reducing the negative experiential impacts of ad overload. Browsing and shopping experiences will become more organically intertwined, and brands will see greater success through thoughtful, authentic, and less intrusive engagement with their audiences.
However, most organizations have significant work to do before achieving this vision. According to McKinsey, only 15% of CMOs feel their company is on the right track with personalization efforts. The remaining 85% recognize the substantial challenges they face, including siloed data, legacy technology, and lack of clear strategy.
Charting Your Path Forward
As AI-native advertising evolves, brands need to develop and articulate a clear vision and point of view for where and how they will apply generative personalization to their advertising campaigns. Before diving into implementation, marketing leaders should:
Consider your overall brand strategy and positioning
Identify the aspects of your marketing strategy that would benefit most from AI
Establish clear boundaries for what their brand will and will not do with AI
Develop a roadmap for building the necessary data, technology, and talent capabilities
Dove's approach offers an instructive case study. As a brand that trades on authenticity and "real beauty," they had the foresight to understand that any AI-generated creative would ultimately undermine their positioning in the market. Dove went so far as to explicitly commit to never using AI in their advertising, recognizing that doing so would contradict their brand essence and market positioning. It’s hard to imagine being the marketing leader that made the tough call to forego the performance gains of generative personalization and cost reduction of AI-generated content to preserve Dove’s brand integrity, but big kudos to them.
The brands that succeed in this new era won't necessarily be those with the most advanced AI capabilities, but rather those who most thoughtfully integrate AI into a coherent brand strategy that delivers genuine value to customers. In the race to embrace AI-native advertising, the winners will be those who use technology to create connections that feel personal without becoming invasive, relevant without feeling creepy, and helpful without being intrusive.
