Everyone is touting hyper-personalization as something that consumers expect from the brands they interact with. Customers expect brands to know their purchase history, their interests, and their browsing history, in order to provide relevant recommendations and suggestions. At some point, however, when hyper-personalization becomes hyper-personal, it becomes downright creepy.

Consumers Want Personalization and Privacy

According to a report from Epsilon, 80% of customers are more likely to purchase from a brand when the brand provides a personalized experience. Additionally, a report from Accenture revealed that 91% of those polled indicated that they are more likely to do business with brands that remember, recognize, and provide them with relevant recommendations and offers. Conversely, research from a Gartner survey on marketing personalization showed that brands are apt to lose 38% of their customers because of poor personalization practices. It’s not that customers don’t want their experiences to be personalized, it’s that the personalization must be relevant, and not overly personal. 

A report from Formation.ai entitled Brand Loyalty 2020: The Need for Hyper-Individualization indicated that 81% of the 2000 consumers that participated in the poll stated that they were willing to provide basic personal information in exchange for a more personalized experience. Of that group, 28% said they were “extremely willing” to provide that information. Those customers expect their data to be used to deliver a top-notch personalized experience from any brand that they have provided their information to. 

Related Article: 5 Drivers of Personalized Experiences: A Walk Through the AI Food Chain

When Is Personalization Creepy?

Each individual has their own perception about when personalization becomes creepy, but the bottom line is, if a customer believes that a brand has crossed the line into creepy territory, then it has. Most people can tell when marketing becomes creepy. 

In 2019 CBSNews reported on a marketing campaign in which women were receiving postcard-style advertisements, congratulating them on their pregnancy — even if they were not pregnant. Even worse, in 2012 Forbes reported that by data mining purchasing habits, Target had determined that a teenage girl was pregnant even before her own mother knew, and sent the girl coupons that would be likely to appeal to a pregnant woman. It’s not that Target’s deduction was wrong — by closely monitoring what a person buys, the brand was able to determine not only that the young woman was likely to be pregnant, but the gender of the baby and approximately when it would be born. This marketing effort, however, most definitely falls into the creepy department. 

While some marketing campaigns obviously cross the line into creepiness, others are borderline creepy. If a customer searches for camping supplies via Google, and the next three sites they visit show them ads for camping supplies, there is a reasonable assumption that the customer may find such personalization intrusive, although many people are tech savvy enough to know about third-party tracking cookies and how they work. If, however, they are simply at home talking with their family or friends about going camping, while they sit next to an Amazon Alexa, and the next three sites they visit show them ads for camping, the marketing campaign has just moved into creepy town. 

The same can be said when one brand sells the demographics and history of its customers to another brand, and the other brand uses that data to market to a prospective customer. The customer is likely to feel that the brand has no appropriate reason for knowing their gender, age, ethnicity, purchasing history, medical history, economic status, etc. A person that recently obtained SNAP food benefits is not likely to appreciate receiving a marketing email from a brand that sells clothes to low-income customers if they have never done business with them brand before. 

The omnichannel nature of marketing today means that customers interact with brands through many devices and mechanisms. A customer may be happy to see that Amazon recognizes them when they visit the Amazon website through both their smartphone and computer. They may appreciate it when they receive emails from Amazon that are directed specifically to them based on their history with the brand. But if they walk into a brick-and-mortar Amazon store for the first time and are greeted by name by the door greeter, they are very likely to be creeped out. 

Related Article: What Do We Mean by Personalization?

The Creepy Scale Is Sliding

Jeanne Jennings, CEO at Email Optimization Shop, said that the scale with which people judge what is creepy and what is not has been sliding, and things that were creepy last year may not be creepy today. “The threshold of creepiness is shifting. Back in 2015 my audience of email marketers overwhelmingly found a ‘browse reminder’ email that I had received ‘creepy.’ It was customized with products I had viewed but not put in my cart on a recent visit to the brand’s website. Just two years later I shared the same email and none of the marketers in the room found it creepy.”

It’s not just that brands know who customers are, but they know where they are. If they are driving by a McDonalds and get a notification that they could receive a Whopper for $1, while they might be delighted at the cost savings, they may also be a bit daunted that McDonalds knew their precise location. Transparency is vital in these situations. If an app or other channel has the ability to track the customer, or if facial recognition is used in a store to determine who you are, your customer history, etc., brands need to ensure that the customer is aware of these “features” before they are used for marketing.

This type of transparency will go a long way towards making the personalized interaction seem acceptable versus being uncomfortable or creepy. Without it, the customer is left wondering how the brand obtained their information. “Will the recipient wonder where we got the personalized information? If it can be clearly traced to an interaction they’ve had with us, that’s usually not a problem,” said Jennings. “Examples here would be past purchases, links clicked in email messages, and browsing behavior on the brand’s own website. If it can’t be clearly traced to an interaction they’ve had with us, that’s a problem. The most common cause of this is data that has been appended to a record, things like location (city, state, zip) and purchase behavior from another brand’s site (even if it’s a sister company).”

Jennings gave an example of creepy marketing that happened to her during an interaction with a brand. “Creepy is bad. A while back I received an email receipt from a large department store. The subject line included my name — not necessarily creepy. But it was creepy because it included my FULL name — the first and last name that I’ve used for nearly 30 years (not creepy) along with my full middle/maiden name, which I have not used since I was married nearly 30 years ago (I’ve used just the first initial). This simple email receipt caused me hours of concern over where they’d gotten my full maiden name; I finally tracked it down to the debit card I had used. It was for a bank account that I had opened using my social security card, which includes my full middle/maiden name. So they must have pulled it from there. But even knowing that it’s still creepy.” 

Learning Opportunities

Related Article: Personalize at Scale With Modular Content

Hyper-Personalization Doesn’t Always Increase ROI

A recent State of Account-Based Marketing (ABM) report from Conversica, indicated that many sales and marketing teams are under-personalizing their communications to prospective customers. According to the report, although these teams want to personalize their efforts, they're facing challenges involving time and capacity to review intent data and tailor each message. The report showed that of those who participated in the 227-person study, only 49% of salespeople and 43% of marketers indicated that they currently use intent data to personalize prospect communications, even though each salesperson is spending 16 hours a week researching prospects for personalization. 

Others are not using hyper-personalized marketing because they do not feel that it will provide an increase in ROI. “Many marketers overdo hyper-personalization because they see it as the key to success and a higher return-on-investment. But this isn’t necessarily true. Even when it’s not creepy, it is not guaranteed to increase ROI. I created a hyper-personalized version of a marketing email promoting toys for a client. The content was hyper-personalized based on the age and gender of the children the email recipient bought gifts for (based on data provided at email sign-up),” said Jennings.

“We A/B split tested this against an email with curated content, a mix of top sellers and new items, which was not personalized. The email without the personalization generated a higher ROI and took less time to produce, since we didn’t need personalized items for each gender/age pairing,” Jennings explained.

Jennings said that the moral of the story is that everything needs to be tested to determine if the efforts that are made actually increase bottom-line performance. “Sending a hyper-personalized email that recipients find creepy can negatively impact the brand; sending a hyper-personalized email that generates a lower ROI than a less- or not-at-all-personalized email is bad for the brand’s bottom line — whether it’s creepy or not,” she explained.

According to Emad Hasan, former Head of Data Analytics at Facebook and Paypal, and current CEO at Retina AI, a customer intelligence solution provider, personalization comes with a cost that many brands do not take into account. “When it comes to customer acquisition, the personalization story also comes at a cost to the business. Personalization may also mean that the business now has to come up with a multitude of creatives that fit each individual. This is a cost that is often incurred by creative marketing teams. Striking the right balance between the number of creative and personalization becomes very important to keep each customer profitable,” he said.

Is Hyper-Personalization on Its Way out?

Others in the industry believe that the end of hyper-personalized marketing will soon be upon us, like Dan Pantelo, CEO and founder of Marpipe, a creative testing platform provider. He thinks the announcement by Apple of iOS14 was a nail in the coffin for advertisers that rely on invasive audience data. “In the post-iOS14 world, hyper-personalization will no longer happen through invasive audience identifiers, but through creative identifiers. Essentially, creative optimization is the new audience targeting,” he said. “Before the advent of iOS 14, advertisers could understand a single customer’s purchase intent based on past browsing and purchase data,” said Pantelo. “For example, if I can see the past five purchases that you made on your Amex card, I can predict with reasonable certainty what your sixth purchase will be, so I already know what you're going to be buying because I'm tracking all of your browsing and purchase data.” 

Now that marketers won’t be able to accurately predict what customers are going to purchase, at least those who use iOS devices, they will be forced to market to a much wider audience. “My ad creative must be more compelling and relevant than ever before to inspire the purchase,” Pantelo explained. “It’s in this way that the updates to iOS 14 will make creatives function more like scientists and engineers, finding out which ads work well and why, so they can replicate those results moving forward.”

Take the Creepiness out of Hyper-Personalization

The key to taking the creepiness out of a hyper-personalized interaction is simply not using information that the brand has no reasonable way of knowing. Hasan said that explainability is vital when it comes to personalized marketing. “The gap between a good recommendation and a creepy recommendation is often filled by explainability. Some of the best recommendations will not only share the recommendation but also suggest why you are seeing that recommendation. For example, you might see ‘people who bought X also bought the following products.’“ 

Chris Hill, CSO at Retina AI, thinks consumers have no problem with hyper-personalization provided that the information that is used by the brand is data that it is reasonable for them to have, such as purchase history. “Customers are looking for a personalized experience based on information they believe is appropriate for a company to know about them,” said Hill. “Recommendations based on past purchase behavior, branded social media interaction, quiz completion or search terms resonate as an authentic personalized experience, or in other words first-party data. However, when brands leverage information that is clearly derived from another source (i.e. third-party) it feels creepy because the company knows something about the consumer that may not have intended or wanted that company to know."

Final Thoughts

Consumers today expect the interactions they have with brands to be personalized for them, rather than people like them. The problem arises when hyper-personalization becomes hyper-personal, and brands use information that is not appropriate for them to have access to. Brands must ensure that their personalization efforts are positively affecting their bottom line, and above all, they must be transparent about how they have obtained the data that they are using.