Building a Persona Menu for Effective Messaging

The most important thing to remember when pursuing holistic personalization is that the effort is about the customers—not the brand, channel or product. If you don’t know much about your customer, how could you pamper them? The trouble is that mounds of data do not readily reveal what individuals are all about. Let’s say a marketing team wants to identify target customers with strong affinities for new product they are about to push. One can imagine a few variables that may intuitively describe imaginary targets, such as high-income families for luxury products. But such rule-based targeting often misses the mark by a wide margin. For one, do you think that the differences between luxury car buyers and high-end TV buyers simply comes down to income? Summarize large amount of data through modeling In a case like this, the best way to identify and summarize all relevant data is to build a model that expresses degrees of desired affinity. Think of it as the level of likelihood that the subject resembles the target. On a scale of 1 to 10—where 10 is the most likely to have a strong affinity—where would a customer fall? Marketers have been employing modeling techniques for ages, mostly for targeting purposes. Out of millions of customers, who should receive high-end home appliance offers? Personalization Is a Team Effort, and Technology Is Only One Player The simplest approach: Among identifiable homeowners in the contact universe, build a model targeting previous responders to similar offers, and rank scores from high to low. Then, contact people with higher scores to maximize response rate, while reducing costs. That would be a classic use-case for models. But models aren’t just for selecting campaign targets. One of the best features of model scores is that they nicely pack all relevant data for the algorithms. In a data rich environment, that is the most important benefit of models, whether they are built by humans or machines. Building a persona menu for your business Personas are nothing more than a series of propensity model scores, expressing an individual’s likelihood of resembling the target. Everyone knows the basic modeling steps, but it's important to recognize that the intentions must be set by the marketers. Exactly what kind of customer affinities do you need for your marketing efforts to succeed? Goal setting is one thing that machines cannot do for you (at least for now). Let’s say you are in travel industry, and you want to pamper your customers with personalized offers and treatments. To do that, you need to know what kind of travelers they are. So, let’s imagine a list of useful personas for you: Frequent FlyerFrequent Business TravelerForeign TravelerLuxury TravelerBudget ConsciousGourmetWine EnthusiastAvid CruiserAdventure SeekerFamily Vacationer These are just a few examples. And the only limitation here is your imagination. You can be totally selfish, emphasizing products and services you would like to promote first, and then imagine types of customers that would be good matches to your offerings. What if you’re in the telecommunications industry? Consider these categories: Home OfficeTech SavvyBig FamilyLandline UserInternational CallerOnline GamerHeavy Movie StreamerAvid Sports Fan (by category such as football, baseball, golf, etc.) People generally respond better when they see something they can relate to. It is about finding something that resonates with them. Brands can do this by gently nudging the answer as to why they might need to upgrade to a faster internet service plan, rather than pushing their agenda indiscriminately, for example. For retail business in general, what types of buyer affinities would you care to know to differentiate offers? Early AdopterBargain SeekerFashionistaLuxury ShopperShopaholicChildren’s InterestOrganic FoodFitness EnthusiastEnvironmentally ConsciousDo-it-yourself HomeownerPlus, any specific brand or product affinities Imagine all these entries are expressed in 1-10, 1-100, or 0-9 scales. To build models like these, all you need are sample targets (such as people who have shown desired affinities in the past). With multiple personas attached to an individual, it would be easy to identify dominant characteristics of a customer by comparing scores side by side. If someone is more of a bargain seeker rather than an early adopter, don’t lead him to the cutting-edge stuff but provide discount offers for more mature products. What now? In this day and age, marketers must act fast before the target individuals lose interest. There is no time to dissect piles of data at the moment of engagement. With a series of prebuilt personas, not just raw data, personalization engines would work better, too. And the result would be far better than relying only on last transaction and immediate browsing data. Consumers aren’t mere extensions of their last actions on your site, and model scores represent vast amount of diverse data used in algorithms. Consider a few persona ideas unique to your business, immediately. If a long list feels burdensome in the beginning, start small with five or fewer personas as test runs. Remember: Lack of imagination is the issue, not lack of data or technology. Personalization Unlocks A New Era of Shopping