The Gist
- Avoid over-personalization. Over-personalization in marketing can harm brand identity, limit customer discovery and reduce overall marketing effectiveness.
- Understand AI limitations. While AI can enhance personalization, it also has limitations and risks, such as data quality issues and over-targeting.
- Prioritize strategic balance. The 1:1 marketing paradigm should be viewed as a guiding principle, not an absolute goal. Marketers must ensure that AI-driven personalization efforts do not overshadow brand messaging or critical business objectives.
Until recently, very few brands have needed to seriously ponder how much personalization is too much. Yes, personalization has been a major marketing trend for nearly two decades.
However, even as brands have gained new and increasingly sophisticated ways of personalizing content, they’ve had their aspirations limited by numerous realities, including their marketing platform’s capabilities, data quality and availability, time and resources and performance visibility limitations.
Now, a lot of people (and startups) are convinced that all limits have been removed. Now, brands can finally achieve their long-standing goal of the 1:1 marketing paradigm — that is, to send the right message to the right person at the right time. And they can achieve that by having generative AI write 1:1 messages for every individual who’s opted in to receive marketing messages via email, SMS and push — and then using machine learning to optimize audiences and timing for every message.
There are some major problems with this thinking. Here’s how over-personalization in marketing can hurt your overall strategy.
The 1:1 Marketing Paradigm Is Aspirational
It’s not a literal or absolute goal. The idea was never for brands to eventually uniquely personalize every message they send, nor was it the idea that 100% of a message needed to be personalized. Personalizing some messages is great. And personalizing part of a message is more than good enough.
Until now, it hasn’t been necessary to specify that. But generative AI has inspired a lot of “can we do this” thinking and not nearly enough “should we do this” thinking. And now when brands think about “sending the right message to the right person at the right time,” they think “right” means hyper-personalized.
Let’s look at some of the limits of the 1:1 marketing paradigm and highlight why over-personalization in marketing can actually harm your performance.
Send the Right Message
Excluding broad, one-size-fits-all messaging in favor of only personalized messaging can potentially cause a number of problems.
Personalization Can Undermine Business Operational Goals
Marketing is a relationship-building endeavor. And relationships involve two parties meeting somewhere in the middle. Yes, marketers want to serve their customers, but they also need to serve their business. Doing the former doesn’t automatically accomplish the latter.
For example, if you have excess inventory in a particular product, you’ll likely want to promote it to more than simply those customers who have bought it before or purchased something else in its product category. Similarly, excess inventory in particular stores should lead you to promote in-store specials or discounts to all of your customers near those stores, not just customers who have demonstrated in-store shopping behaviors.
For new product launches, you’ll likely promote it more heavily to customers who have shopped that product’s category, but it would be a missed opportunity if you didn’t promote it more widely than that.
Personalization Can Limit Discovery and Inspiration
If you only show your customers stuff that’s similar to what they’ve already expressed interest in, you limit their growth as a customer. You essentially put them in a box.
For example, I recently learned about a nonprofit that sent an email campaign highlighting programs the recipient could consider supporting, but only highlighted groups that were in the same category that each subscriber had donated to in the past. So, if you donated to an educational charity, then you only saw education charities as options, depriving them of the opportunity to explore charities with other missions and thereby reducing donations overall.
Personalization Shouldn’t Crowd Out Brand Messaging
What are your brand’s values? What kind of image are you trying to cultivate? Especially among those who are younger, consumers want to know what your company stands for. This kind of positioning shouldn’t be personalized, nor should it be neglected in favor of more personalized content.
We recently had a client that pulled back significantly on AI-driven personalization for this very reason. It was crowding out brand messaging, and they know brand-building is important.
Brands Want to Create a Common Experience
Most brands don’t want each of their customers to have completely bespoke experiences, as that can undermine brand identity. We work with lots of enterprise clients, and they all have rigorous and detailed brand guidelines that cover brand voice, acceptable imagery, design parameters and more. Even a generative AI system that’s trained on those guidelines wouldn’t be able to create consistent enough experience to satisfy most enterprises.
And with highly regulated industries — like pharmaceuticals, government and insurance — letting generative AI create copy that publishes without human review is a non-starter.
There Are Diminishing Returns on Content Personalization
There’s no shortage of things you can personalize about a message. In fact, we identified more than 170 criteria in our Segmentation & Personalization Ideas checklist (free, no-form download). But the truth is that only a few will really make a big difference for any particular message. The rest will do little to nothing — or even hurt performance by being a distraction.
View all
It’s also worth noting that as brands do more nth degree content personalization using generative AI, measuring the impact of all of those variations becomes infinitely more complex to track and gain insights from across audience segments.
Related Article: Generative AI for Email Personalization: A Hallucination Wrapped in Confusion
To the Right Person
Over-targeting using AI is also a significant risk.
Audiences Can Become Too Targeted
Segmentation and suppression rules are critical tools in maintaining an engaged audience and minimizing list churn, since irrelevant and overly frequent messages are perennially top drivers of opt-outs. However, brands need to watch the algorithms used by their machine learning models when it comes to audience selection so they don’t become overly picky.
For instance, I recently heard about a brand that was using machine learning to reduce churn. To accomplish its goal, the algorithm was aggressively suppressing emails to new subscribers, since new subscribers churn at much higher rates than established subscribers. That had a significantly negative impact on sales. The experience caused the brand to realize they needed different rules for new subscribers and existing ones, because the behaviors of those two groups are quite different.
I’ve also heard of instances where brands leaned too heavily into segmented campaigns, where they only sent messages to subscribers who had previously shown an interest in the campaign’s topic. As a result, many of their new and less engaged subscribers received too few campaigns, depriving them of opportunities to engage, which caused the percentage of subscribers who were inactive to increase.
Related Article: 7 Factors That Determine Email Deliverability
Engagement-Based Targeting Isn’t Perfect
That’s because engagement data isn’t perfect. Privacy restrictions like Apple’s Mail Privacy Protection and Link Tracking Protection make some engagement signals less reliable or plentiful than they’ve been in the past. And most brands haven’t centralized their customer data in a customer data platform or similar system. Instead, customer data is strewn across channels and systems, giving them an imperfect picture of whether a subscriber is engaging across channels.
Given the realities of data quality and availability at most brands, data-driven targeting needs to be more forgiving. Having exacting algorithms only makes sense if you have exact data.
Related Article: 4 Ways Brands Go Wrong With Digital Marketing Metrics
At the Right Time
The best time for a marketing message to arrive to an individual subscriber may not be sensible for brands to respect in some circumstances.
Send Time Optimization Doesn’t Always Make Sense Operationally
A subscriber’s historical engagement time patterns aren’t meaningful when they clash with operational realities, such as:
- Wanting to drive in-store or restaurant visits during operating hours
- Wanting to drive call center inquiries during call center hours
- The need to release products or tickets at set times determined by vendors
- Breaking news, the results of sporting events (e.g., Super Bowl), etc.
- Promoting a flash sale
- Promoting doorbusters on Black Friday or other occasions
Again, the “right” time for a campaign isn’t purely decided by the subscriber. The “right” time for the brand also matters — sometimes a lot.
Multiple Campaigns a Day Makes it Moot
If you’re sending multiple times a day, send time optimization typically doesn’t move the needle on performance much. For our clients that run two or three campaigns a day, they typically just space the sends out across business operating hours.
Related Article: How to Prioritize Email Personalization’s Perennially Moving Target
Know Your Limits: Avoid Over-Personalization in Marketing
All of this isn’t to say that personalization and AI aren’t good. They are. But like most everything, they can be overdone.
Remember that the 1:1 marketing paradigm is intended as a North Star to guide you in the right direction. It was never intended to inspire you to launch yourself into outer space. So, stay grounded in your use of AI.
Learn how you can join our contributor community.
