Photo: @criene via Twenty20
What does artificial intelligence mean for loyalty marketing?
Through a special arrangement, presented here for discussion is a summary of a current article from Bill Hanifin, managing editor, Wise Marketer, first published on Forbes.com.
In our experience, the one-to-one relationship aspect of true loyalty marketing isn’t possible to create at scale, nor even sustainable at present levels, without the disciplines and tools that are central to artificial intelligence (AI) and machine learning.
At its most basic level, AI can automate simple, repeatable tasks and accelerate pattern recognition. This opens the door to marketing efficiency and hastens the fulfillment of the promise to provide “the right offer to the right person at the right time, in the right channel.”
It also delivers personalization, meaning practical ways that a brand can demonstrate that it is listening to its customers across their digital footprints (website visits, purchases, searches, posts, reviews, etc.), learning more about what they want and formulating offers for products they actually want instead of what the brand wants to sell. This is the essence of how data-driven marketing can build trust with customers and long-term loyalty.
Enhancing your analytics engine with AI can make it possible to optimize individual experiences at scale. Next-generation analytics, fueled by AI, can make hyper-segmentation possible (i.e., homing in on more precise groups of customers who share specific attributes and behaviors).
As a next step, dynamic content can be optimized for delivery across channels in a timely manner, with the result being a delightful customer experience and higher satisfaction levels.
Customers are becoming more aware of the value of their data every day, and what it represents to the brands they patronize. Delivering a highly personalized experience communicates clearly to the customer that their favorite brand “gets it,” meaning they might just wonder how the brand managed to connect with them in a natural, organic way that encouraged them to visit, purchase, recommend it, etc.
Loyalty marketing is focused on creating long-term, mutually beneficial relationships that transcend the era of segmentation-driven campaign management.
The next time a discussion of AI breaks out in a meeting, don’t avoid the topic due to the unknowns, but embrace it through a belief that it could spark the ultimate revolution in marketing analytics at a time when we seem to need it most.
Discussion Questions
DISCUSSION QUESTIONS: How confident are you that artificial intelligence will be integral to bringing personalization at scale to loyalty marketing? Is the technology showing some limitations or unknowns that cloud its potential?
Poll
BrainTrust
Georganne Bender
Principal, KIZER & BENDER Speaking
Evan Snively
Director of Planning & Loyalty, Moosylvania
The author confuses artificial intelligence with automation, a common mistake. Automation is certainly integral to efficient loyalty marketing on a large scale. Whether it is effective is a different question and the marketing literature is mixed on that question. But nowhere is AI integral to this process – a good modeler can do just as well, potentially better. Then you build the automation around the model, checking back to make sure the model hasn’t gone astray.
Most brands and retailers haven’t even approached 1 percent of AI’s potential.
I agree with Bill’s statement “Enhancing your analytics engine with AI can make it possible to optimize individual experiences at scale” but… that is only a first step and will fall short of meaningful personalization in the long run.
The biggest blunder I see everywhere AI is employed in a consumer/user-facing scenario is keeping it under the covers away from users. AI is most often employed as a secret analysis mechanism which limits it to being only as good as underlying algorithms and the assumptions driving them. There is a complete fear of using those assumptions as a basis of asking consumers/users if the AI-generated conclusions are correct or if not, where they are wrong or what their preferences are. Doing as I describe alongside other queries strategically placed (like in a product discovery phase) greatly enhances the USABLE data that can drive AI and its subsequent personalization.
AI will be integral to personalization at scale. It already is at a basic level. It is only getting better. Currently, at a high level, AI makes mistakes. At lower levels it’s almost 100 percent. As the technology improves, the margin for error will shrink to a point where a brand can, with confidence, use AI to talk to customers “one-to-one.” We’ll move from personas to a truly personalized experience.
I have never had a truly personalized experience with any chain, big box or grocery store. The closest thing to it is a general email that leads with ”Dear Georganne” or, more frequently, “Dear Bender.” Truly personalized experiences are still made through a store associate who takes the time to tell me about an event, an item I might like, or who sends a handwritten note to say thank you for a purchase.
If AI can help retailers connect with shoppers in what appears to be a natural and organic communication then I am all for it. I just haven’t seen it yet.
A lot of the attraction of artificial intelligence is the belief that it removes both the vulnerabilities and practical boundaries of human involvement – as if it were some kind of auto-pilot. The truth is that artificial intelligence is only ever as good as the human intelligence behind it. This requires a deeper understanding of the human experience of marketing, of being a consumer. Such a commitment is a departure from the prevailing wisdom that created today’s most prolific data-driven tools – like programmatic advertising, retargeting and other experiences of marketing that have frustrated, more than aided, consumers.
You nailed it, Susan!
Artificial Intelligence (AI) can and will be key to delivering personalized customer journeys. The challenge as mentioned above is scaling it effectively across channels. It is almost impossible to achieve cross-channel personalization without eliminating the separate silos of automation that exist in almost every retail organization. Amazon does it well in one channel (web) because it is centralized, cloud based and services a single channel.
The store must move to the cloud to allow scaling AI across the in-store and web channels that drive today’s omnichannel loyalty marketing.
Personalization at scale is a sound bite mirrored against the lackluster results over the past 25 years as retailers and marketers invested millions into “personalization” schemes and bright shiny technologies. The #1 big, messy problem? Digital is devoid of sensory cues. The #2 big, messy problem? Computer science vs. human cognition. Decision trees vs. human emotion driven by individual sensory preferences. Sensory failure, rendering current personalization technologies defective at scale. Al leveraging machine learning must know what to solve for. True one-to-one human “personalization” is dependent upon an individual’s unique sensory preferences. Proof personalization has failed? 3 percent online conversion and 35 percent to 50 percent return rates.
In fact, reverse logistics are consuming ROI at a rate heretofore never experienced by retailers. And the reason millions of customers return goods? The reason for returns at Amazon: 34 percent said the size, fit, or the color was wrong, 14 percent said item not as described, 10 percent disliked item, 9 percent changed their mind, and 4 percent intended to return the item anyway, as many buy two or three of the same item to buy one matching their sensory preferences of fit, look and feel.
Today is the beginning of true personalization. Retailers and marketers need to evaluate “personalization” strategies and tech from a human perspective. The new personalization technologies must leverage AI to elicit individual human sensory preference intelligence. This is the elephant in the room squashed for years by the endless dichotomy of using finite, linear mathematical formulas to solve for individual human-based cognition. Explicit vs. implicit human behavior, the dichotomy of computer science and human-centric abstraction. Think about it all you individual humans who have read this far, after all, you may or may not have preferred to read my comments.
As always, technology is ever-evolving, and AI is no exception. For instance, I am a part of a natural voice technology pilot, and it has a bit of improvement to go. However, I see this particular pilot showing immense progress and huge potential in the marketplace. I am extremely confidant that AI will help relevant personalization, as it is and has been already for several online retailers.
Increasing efficiency (the only real opportunity in AI) is never the way to long term success — but it’s one of the important things to do along the way to avoid failure.
Somehow, business culture has become so wrapped up pursuing efficiency that we’ve forgotten what customers really want: Opportunity to buy the products we want in a well structured store/online store, at good value, and supported by good customer service.
AI can play only a small part in this. Unfortunately, it appears that the process of implementing AI systems is so difficult that retailers might lose years of strategic development getting their AI in place. None of us can let that happen.
AI is a smart direction, but won’t deliver long term strategic value.
Bill provides a great perspective on loyalty and the power of leveraging data. What I really like is he never loses sight of the goal being to serve the customer and improve the brand or retailer’s relationship with the customer.
AI and machine learning can be powerful tools to personalize the experience for shoppers and make them feel known and understood. However too often, the technology is seen as a way to predict what customers will do, so we can manipulate them to serve our needs. That’s when it can get creepy and turn customers off.
It is refreshing to see an article on AI that includes discussion of the impact on the customer. Bill reminds us that the long term goal of loyalty marketing is “to establish valuable, long-lasting, mutually beneficial relationships that insulate the brand from competitive offers.”
The answer all depends on your timeframe. If we are talking now until 2025, I’m not very confident. Beyond that, anything is possible. The author says AI can automate simple repeatable tasks. Well, maybe in the broadest sense of the loose definitions, but by those same definitions so could a robot. The author also says that AI has the capacity for pattern recognition and, again, in the broadest and least accurate sense of the terms, he’s right. But there is a huge difference between recognizing patterns and recognizing the correct patterns just as there is a chasm of difference between pattern recognition and pattern imposition. Amazon uses algorithms to recognize patterns, but the results are often mind numbing. For example, I once bought a book on Beat novelist William Burroughs, who was gay. Amazon began sending me all kinds of recommendations based on my “pattern” of liking LGBTQ books. Weirder still, their first recommendation was Friedrich Nietzsche’s “The Gay Science” — or “Die fröhliche Wissenschaft” for you German readers out their who might more correctly translate it as, “The Joyful Wisdom” an 1882 book that has nothing to do with being gay. And that isn’t beginning to address the problems current pattern recognition systems have recognizing the “why” behind the “what” of purchase history and/or changes in buying behaviors. Will these issues be worked out in time? I believe they will, but that time is still a ways down the road to technological Utopia.
I’m not sure it’s the technology that is showing limitations, but more commonly it is the practitioners lacking imagination on how to best leverage it. Part of that comes from treading into new waters and not fully knowing what could be possible, but one thing is for sure – that problem won’t be solved by sitting out on the sidelines.
Loyalty marketing, with its long-term orientation, is the perfect sphere for utilizing AI.
On the topic of achieving true 1:1 personalization, a major goal of loyalty marketing is helping customers feel part of the brand community – so while many brands today are losing sleep over the arduous task of customizing every touchpoint, they must not lose sight that it is still critical that they create an environment where customers can say to brands “I like what you are.” Not what you are trying to be to me, but what you are. When that identification happens consumers will respond by actually personalizing themselves to the brand, and a loyalty marketer’s job becomes just little bit easier.
We too often see retailers treating loyalty and personalization as separate, at best loosely coordinated, initiatives. They should be deeply integrated. Personalization and the AI behind it feeds on the richness of our data set at the individual consumer level. The #1 way to build this data set is through an effective loyalty program that incentivizes every transaction to be “logged in” and uses progressive profiling to build the non-transactional elements of the consumer’s profile.
Increased personalization is coming. The question is whether consumers will be delighted. Undoubtedly, some will. But I bet many will find it creepy and invasive. Further, there are sure to be unintended consequences … such as the case of the father discovering his daughter was pregnant by mail promotions sent to her.
While it is true that in the future artificial intelligence will play a role in delivering marketing personalization, it is important to recognize that the capabilities described in this article have been available through traditional modeling and marketing technology for a number of years. The strategy of bringing the “right offers on the right products to the right customers through the right channels at the right time” has been present in marketing for many years; however, customization has been available only to this point in segmentation form.
Now, with the introduction of more powerful processing capability, that segmentation can be delivered to smaller and smaller groups, getting closer to the end goal of personalizing communication at a 1:1 level.
Even that level of personalization has been done for limited use cases over the past 10+ years, e.g. Kroger/Sainsbury coupon personalization.
AI will not introduce any new strategies at the beginning; rather the promise is to increase the level of personalization — an important goal, but not a revolutionary one.
People are difficult to predict, things are easier. I think AI machine learning can be used to derive pattern data for loyalty programs, but right now it takes human to understand the nuances and emotions that comes with certain customer interactions and offers. AI is a better fit in the short term in areas like supply chain and merchandising I think where there is rich data, less privacy issues and very little emotions involved.