So Much More Than Coupons: Generative AI and Personalization for Retail-CPG.

A discussion of the use of generative artificial intelligence in the retail-consumer goods industry inevitably leads to the topic of personalization.

Defined, often, as content and advertising – and, in time, products and services – tailored to a cohort of one.

The economic value of personalization – or, better understood as customer intimacy (see below) – is significant.? A 2021 study from McKinsey found that skillful personalization – i.e., careful analysis of data that increases a brand’s customer knowledge, and smart messaging that shapes behavior -- can drive revenue lifts of up to 15 percent.?

McKinsey noted that more than three-quarters (76 percent) of consumers reported that receiving personalized communication was a key factor in prompting consideration of a brand, and 78 percent said such content made them more likely to repurchase.?

A more recent (February 2024) report from Marigold found that 91% of those surveyed said that brands needed to treat them like an individual to win favorite or preferred brand status.

?

The data shows that it’s two-to-three times easier to sell to an existing customer than a new prospect.? And that the longer customers buy from your business, the higher their average order will be.? And that customers who profess loyalty will buy more and more often and across all channels.

Given the value of customer retention in this extraordinarily competitive era – and now, the content creation capabilities of generative artificial intelligence – let us explore, briefly, the what, how, and what’s next of personalization in this time of AI.

?

As we know, the pandemic – with its supply chain pains and closures – sent fissures through shopper behavior.? The 2021 McKinsey study said that roughly 75 percent of U.S. consumers had switched to a new store, product, or buying method in the prior year.?

I suspect that the pandemic’s shock to the system (I can’t trust my retailer as I used to) -- and more importantly, the ever-present flood of algorithmically-driven, personally-curated content from social media providers – has significantly raised the bar of shopper expectations.

And raised it well beyond – dare it be said -- the realm of traditional, reward-based loyalty programs.?

The data shows that seven of ten Americans considered loyalty programs to be leading factor in securing adherence to favorite brands, and that a 2022 survey showed that more than 50 percent of U.S. shoppers were likely to increase their participation in such schemes.

But we also know that six of ten U.S. shoppers want discounts in return for joining a loyalty program, and that loyalty rewards program membership and usage are two different things.? On average, U.S. consumers belong to 16.6 programs – but actively use only half of them.

Yes, coupons that reflect last week’s shopping are nice.? (Yesterday, my wife and I saved $1.75!)

But mind you, the Marigold survey respondents said they wanted to be treated as individuals.

Which, I would suggest, is different from receiving offers, coupons or rewards reflective of past shopping.

?

The 2021 McKinsey report – researched and written in the midst of pandemic -- noted that consumers responded most positively when “brands demonstrated an investment in the relationship, and not just a transaction.”? ?What does that look like?? Cited in the 2021 study were such actions as post-purchase check-ins, the sending of a how-to video, and requesting post-shopping feedback or a review.

All of which are different from receiving offers, coupons, or rewards reflective of past shopping.

Another McKinsey study – one that should be studied by every technologist who utters the word “personalization” in a NRF booth or Shop.org panel discussion – asked customers what they most valued in personalized marketing.? Four big ideas emerged:

1.?????? “Give me relevant recommendations I wouldn’t have thought of myself.”? This is less about reminding shoppers of what they looked at, but didn’t buy, and much more about go-withs – the products and services that complement a completed or contemplated purchase.

2.?????? “Talk to me when I’m in a shopping mode.”? Personalization is also about timing – it’s just as important as content.? Perfecting this requires a close (and constantly revised) look at customer behaviors and habits.

3.?????? “Remind me of things I want to know but not might be keeping track of.”? ?Brands can track events and circumstances – birthdays, anniversaries, wedding season, potential reunions.? (Might the class of ’99 be hunting for cocktail dresses this year?)? ?Brands can also suggest add-ons – freshening last year’s look or maintaining last year’s project.?

4.?????? “Know me no matter where I interact with you.”? ?We’ve known if for more than a decade: consumers care not a whit about channels.? They shop with a brand.? Sometimes in the store, sometimes online, sometimes both.?

None of these four ideas are about offers, coupons, or rewards reflective of past shopping.

Perhaps there’s a gap – a big gap -- between what they’re getting and what they’d really value.

?

I suspect that development and delivery of McKinsey’s four would lead to the creation of a known, trusted relationship between a brand and a consumer.? Former Cisco colleague and customer experience expert Rachael McBrearty described this as customer intimacy – a relationship a step beyond transaction and interaction, a relationship built of deep knowledge, a relationship that might resemble that of caring siblings or best friends, where interests, dreams, and stories can be freely shared without fear.??

Intimacy is about trust.? It’s also about confidence – note the phrase “freely shared without fear.” It’s comment, a gesture, a purchase made without hesitation or remorse. Esteemed Indiana University Professor of Marketing Dr. Raymond Burke has found in his “Shoppability” research the importance of retailers lowering decision anxiety and increasing confidence.

Hesitant? An abandoned cart. Assured? Thank you for your order!

?

Might brand-shopper intimacy – built of knowledge, trust, consumer confidence –be possible at scale in this new AI era?? ?Might it become the new reward of a loyalty program?

?

If so, how do you get there from here? ?

Start not with generative AI content, but with a sharp-edged analysis of your data. How much do you truly know of your customers? Is data cohesive, clean and accessible across the company? What about external and unstructured data that may shape demand or spark ideas? What policies and safeguards permit or prohibit??

The recommendations, timing, and reminders desired by shoppers are possible only when a brand knows its shoppers – and that that which influences their consumption decisions. ?

Customer-behavioral-trend data is the ammunition of the now-underway retail AI war. More is generally better than less. This includes internal and external data, structured and unstructured data.

Those who have it will win shopper attention, loyalty, and revenue. Those who don’t will wonder why.?

Start not with generative AI content, but with a sharp-edged analysis – most likely through AI deep learning -- of your customer cohorts.? How defined? How small, how precise?? Given an expansion of data availability (especially of your most loyal customers) and deeper analysis, what might be the cohort-specific triggers for a next shopping action, a next step on the decision journey? ?

Start not with generative AI content, but with a sharp-edged discussion of customer trust. What creates it, and where (and how quickly) it can be lost. How is personally identifiable information respected and protected? Why should a customer trust your brand with even more secrets??

Again, we’re talking about customer intimacy – not transactional coupons.?

Start not with generative AI content, but with a sharp-edged evaluation of the trustworthiness of your generative AI language model. Will its outputs reflect brand truth and brand style all the time??

?

When those four steps are in the rearview mirror, turn to generative AI.

Yes, it can generate personalized content. However, the question is not the personalization, nor the AI technology, but the content, and its value to the shopper.

Let’s examine opportunities for McKinsey’s first finding:? give the shopper relevant recommendations I wouldn’t have thought of myself. ?This is an area of tremendous opportunity, given generative AI’s ability to create, summarize, and personalize.

A recent white-boarding session led to these ideas. None are necessarily new.? What’s new is our ability – thanks to generative AI – to produce them personally and at scale.

o?? Go-with suggestions.? Accessories and complementary items to the purchase. It’s more than “other customers also bought this;” the effective personalization (based upon far-sighted data acquisition and preparation) references the customer’s recent-past purchases, likes and pins, the anniversaries and special events.? These are suggestions from a best friend or expert personal shopper, one who knows your kids and their birthdays – and inspired not just by existing inventory, but the wide world of trend and design.

o?? Recipes. And not just for meals and beverages. If a recipe is a list of ingredients and step-by-step instructions, then recipes can be written for the building of decks, the growing of green lawns, and the installation of home Wi-Fi.??

In the food and beverage category, it’s a given.? Recipes sell. A 2023 survey indicated that 88 percent of U.S. consumers use guidance from food blogs, recipe sites, and social media or video platforms (Tik Tok and YouTube, primarily); interest and use was highest with younger consumers, with roughly 90 percent of millennials and Gen Z cooks using online recipes.? Most important:? about three of four consumers surveyed said that an online recipe had inspired them to purchase an ingredient or product not previously purchased. Grocers:? keep in mind that food and beverage site allrecipes.com reported roughly 146.9 million visits in February 2024.? Not bad.

Personalization? How about a festive holiday dinner menu (with ingredients and prep timing-to-table) for 15 with a guest list that includes one vegan, two vegetarians, one with gluten sensitivity, and a type II diabetic?? How about the exact list of lumber, gravel-soil, and mechanical fasteners needed for a multi-level raised rose and flower beds that faces north in the northern hemisphere?

o?? How-to-dos. This is the strategic sibling of recipes, but at a higher level. It’s the personalized answer to the plea of “how do I get great internet connectivity all through my house?” or “how do I change to a new cellular carrier without losing all my contacts and data?”? It’s guidance as to how EV drivers can travel from Portland to the big game in Eugene on Saturday without wandering in search of charging stations – or spending hours waiting for a charge.?

It's addressing, in a personal way, the barriers to purchase, the lack of knowledge and confidence that stands in the way.

o?? Summations of reviews and ratings.? ?It’s not what everyone says, it’s what people like me say – the peers of my cohort.? It’s about what we hold dear.??

Don’t make me scroll through hundreds, if not thousands, of comments.? Give me a concise summation, pulled not only from your site, but the publicly available sites of others, even professional reviewers.

Address, in a personal way, the barriers to purchase, the lack of knowledge and confidence that stands in my way.

o?? Summations of product comparisons. How, exactly, does product A match up to product B?? Or C?? How about a concise summation of not only the specifications and price, but (see above), the consensus peer evaluation??

There’s much more generative AI can do in the service of personalization – which is in the service of customer retention and loyalty.

Of course, generative AI can translate content into easier-to-read material or turn it into audio or video, knowing that some 54 percent of American adults lack “literacy proficiency” – the ability to understand texts beyond filling out basic forms, or to correctly evaluate the reliability of written content.

Of course, generative AI can translate content into the language of comfort and fluency, knowing that, according to the U.S. Census Bureau, more than one in five Americans does not speak English at home.

Of course, generative AI – in concert with natural language processing and generation – can share personal, well-timed (and highly desired) information on order status and delivery.

Of course, generative AI – in concert with natural language processing and generation – will deliver to us personal shopping agents, conversational bots that guide us through shopping sites, answer questions, offer insights-reviews-comparisons, and provide assurance of a good decision.? And, in time, such a bot will represent each one of us in the marketplace, seeking best possible outcomes on our behalf 24-7.

And, of course, generative AI can address the other three McKinsey findings for personalization.

Those are topics we’ll turn to in future postings. ?

?

Your comments and criticisms, please.

?

I’m Jon Stine, 35+ years in retail business and technology.? Most recently in conversational AI.

I read, I listen, I observe. I think, I write, I advise.

[email protected] , +1 503 449 4628.

?

This article also published at j.christopherccv.com .