Of all things that keep CMOs up at night, data is the worst culprit. More CMOs felt unprepared for the "data explosion" more than any other marketing challenge, according to an IBM Global CMO Study. Under pressure to "personalize" marketing, CMOs now toss and turn at night thinking about the data that should allow them to target relevant content to the right person via the right channel.
Here's the problem: Scale and relevancy don't mix. The more people marketers try to reach—the more consumer segments they carve from data—the harder it becomes to deliver relevant content to each person.
So how far should CMOs take personalization? At what point do the extra creatives, ad placements, and billable hours stop delivering a return on investment?
To personalize on a scale that makes sense, CMOs need to look at the blind spots in data that can cause collisions between scale and relevancy. CMOs must also look at the nature of each brand and product to determine whether personalized marketing is worth the investment.
Social Media
Social media is one of the most promising avenues to personalization, but it has a blind spot that can wreak havoc at scale: "dark social."
Alexis Madrigal, contributing editor at The Atlantic, coined the term "dark social" to refer to all the social sharing that can't be measured by Web analytics. In text messages, emails, Skype conversations, Facebook messages, and other private channels, people talk about products, and brands can't listen in.
Dark social accounts for 69% of all digital sharing, according to Madrigal's calculation and a more recent study by the ad platform RadiumOne. That means that public data on Facebook, Twitter, and other networks is less than a third of the conversation. Over-reliance on this data can lead marketers to misjudge their audience and scale irrelevant content.
If you read reviews, company Facebook pages, and social comment sections, you see extremes: delight or outrage. People with tempered opinions are either silent or express their views via dark social. So when brands examine social media data to understand public sentiment, they only see the fringes.
Given the volume of social data, most brands have to use language-processing technology to identify consumer segments and analyze their sentiments.
Unfortunately, this technology struggles with context, lingo, sarcasm, and other nuances of communication. For example, let's say I go to a brand's Facebook page and post, "Awesome job customer service!!! Thanks for keeping me on hold for 70 minutes!" The language processor could interpret that as a positive sentiment towards customer service instead of sarcasm. We human beings have a hard enough time interpreting each other's text messages and IMs; we can't expect computers to be better at it (not yet, anyway).
Consider the potential impact of this blind spot. A marketer wants to study social media data to identify consumer segments with distinct views towards the brand and its products. However, the majority of social sharing is invisible, and the visible remainder contains the most extreme views. Natural language processing technology is likely to misinterpret some of those views. Thus, the marketer could misunderstand the audience and then spend money creating and distributing irrelevant content. At scale, that could be a costly mistake.
A Considerable Challenge
In the conflict between scale and relevancy, the most well-known exception is Amazon. Every Amazon purchase can be tracked; there is no "dark" shopping. The company can personalize marketing to each individual shopper. However, most brands—especially those in fields like professional services, healthcare, and nonprofits—have "light" and "dark" interactions with customers.
If such brands could segment consumers based on the available data, would personalized marketing be worth the cost? The nature of the product is often a better guide than the availability of quality data.
Take smartphones as an example. Some consumers conduct research, ask for opinions, and consider multiple options when choosing a smartphone. They treat smartphones as a "considered" product, similar to how they would treat a car, college education, or B2B software. Some people choose phones emotionally–they have an affinity for Apple or Samsung that supersedes the practical differences between their devices. They treat smartphones as an "unconsidered" product, much like way they would treat Coke versus Pepsi or Brawny versus Bounty. No one asks for a friend's advice when choosing between these brands.
With the "considered" group, personalized marketing could pay off. If Samsung or Apple could determine what features excite a specific segment of consumers, then yes, developing content for that segment might increase sales.
With the "unconsidered" group, personalized content probably wouldn't be as worthwhile. Generic, emotional ads would have as much or more impact than considered ads. Thus, for most brands, it makes sense to scale relevancy to the "considered" groups but not the "unconsidered" segments.
The CMO's Struggle
Personalization has been put on a pedestal, and each CMO must determine where the quest for relevancy will clash with scale. Acknowledging the blind spots in data, which consumer segments are reliable? Which segments merit personalized marketing given the consideration or lack thereof that goes into buying the product? That is the CMO's struggle.
Most companies can't get air-tight data like Amazon can, but that does not make data unusable.
* * *
We shouldn't ignore social media or summarily throw out personalization. Rather, when CMOs try to mix relevancy and scale, they have to be critical of their data and how it's used. And yes, that might cause a few sleepless nights.
