Image: Kroger
How important is non-transactional data to personalized offers?
Denise Leathers
Through a special arrangement, presented here for discussion is an excerpt of a current article from Frozen & Refrigerated Buyer magazine.
Kroger uses an abundance of transaction history from its loyalty cards to create personalized offers for its customers. But it also taps a whole host of lifestyle attributes based on demographic, geographic, psychographic and other external information to further refine those offers.
“Our mission is to making people’s lives easier by harnessing the data to create a more personalized, more relevant in-store experience,” said Yael Cosset, chief commercial officer and CIO of Kroger’s data analytics and marketing firm 84.51. The service grew out of Kroger’s long-time partnership with U.K.-based customer science company dunnhumby USA, whose technology assets it acquired in 2015. “But to truly understand what matters to our customers, we need to go well beyond the traditional way of looking only at the transaction … [and include] what happens around it as well,” he explains.
So, why did customers buy a particular product? What did they think about it beforehand? Where did they get their information? Who or what influenced the purchase? Said Mr. Cosset, “All of that feeds into our knowledge of the consumer and what matters to him or her.”
Today, thanks to the digital revolution — and mobile phones, in particular — Kroger finds itself able to gather a lot more customer information than even it may have thought possible.
“The important change that’s happened over the past six years is that the consumer’s path to purchase used to happen inside the four walls of the store, and it was all observable,” says Brian Kilcourse, managing partner at RSR Research. “Now the path to purchase begins outside the store, in the digital space. But it’s just as important to observe those non-transactional behaviors. It’s harder because it takes technology to do it, but retailers have to in order to remain relevant to consumers.”
Discussion Questions
DISCUSSION QUESTIONS: Do you see non-transactional data becoming as important as transactional data in guiding personalized offers to consumers? What are the inherent challenges in using non-transactional data to tailor offers?
Poll
BrainTrust
Ken Cassar
Principal, Cassarco Strategy & Analytic Consultants
William Hogben
CEO, FutureProof Retail
Non-transactional data collected is complementary to, but a replacement for, transactional data. There is much value in knowing things about one’s customers such as geographic location, size of household, household income range, preferences at other sites, etc. But these data only complement the knowledge of what a customer bought or didn’t buy (market-based analysis), or when purchases by a customer are the largest. When combined with non-transactional data, a retailer can further analyze what drives the purchase of certain products at certain times via certain channels, and that is extremely helpful in defining the holistic approach to marketing to individuals.
I agree with Bob and would further add that for customer-centric companies, the additional clarity gained with non-transactional data can be substantial. But this assumes that data is being collected and used versus the vanity metrics of simply collecting data. Hence truly customer-centric companies have the advantage because they tend to ask relevant questions and derive actionable insights as a result.
Non-transactional data (NTD) is rarely as important as transactional data (TD), but it can help add additional insights — assuming that the TD is fully leveraged.
Too many retailers believe that the sheer quantity of NTD can compensate for the greater quality of TD — but that’s almost never true. NTD may seem to be more interesting, actionable and insightful than TD, but it’s not nearly as predictive (which is what matters first and foremost).
Retailers must walk (with TD) before they attempt to run (with NTD). Here’s a recent interview that picks up on this same theme: “Big Data: Start with Baby Steps”.
While transactional data is incredibly important as fuel for targeted promotions and offers, it has limitations. If, for example, I know that someone bought a flat panel TV set, it’s too late for me to try to influence that decision. One of the really compelling opportunities that the online space opens up is the ability to target ads one-to-one to people based upon intent signals. Data from Foursquare might tell me that they visited an electronics store and I might’ve seen that the same person visited the manufacturer’s website to review specifications of a particular model but did not purchase. Signals like these, while they’re not bright neon signs saying that a person is about to buy your product or a competitor’s, ought to be a part of marketing strategy for every high-consideration product manufacturer or retailer.
Non-transactional data helps complete the story and that is key. Today’s consumer wants to be understood and interacted with, not sold to. They want you to serve them.
To do that, retailers must understand who that customer is and what they might want from the brand — that goes beyond a product.
You know my motto: Serve before you sell … for my 2 cents.
While I certainly agree with Professor Fader that transactional data is the most powerful tool with which to personalize offers, limiting our personalization only to those with transactional data leaves us presenting generic offers to non-customers (who vastly outnumber your current customers) and any customer prior to being “identified and connected” to their transaction history.
We’ve had great success leveraging non-transactional data to tailor online content and offers for non-customers and unidentified customers. These “personalizations” regularly boost online conversion by double and even triple digits.
Examples include using the referring domain to present a competitive offer, using the geo-IP to present a locally-relevant offer and using a cookie to present an offer in response to browsing history. These and many other similar campaigns have generated large conversion gains in the most common scenario for retailers (i.e., they are presenting a user experience to a visitor for whom they have no transactional history available).
According to retailers we’re working with, there is one major source of non-transactional data they, to date, been unable to capture and use. But that’s changing. In observing in-store shoppers on mobile, they believe that as many or more shoppers are capturing data for future probable purchases versus showrooming. By that I mean the consumer has researched a considered purchase online and now is in store to see it, maybe get a question answered, and compare it to nearby SKUs. They then take a picture of the item/price card, and email/text/social share the image to themselves. The retailer of course captures none of this data. But now retailers are getting smart about capturing this dataset for use in marketing and are embedding “Remind Me” options in mobile product pages where they can alert the consumer on the product, price drops and other.
Non transactional data is the majority of all data — and so of course it’s important to good personalization. Many commenters here are taking an older view of non-transaction data, that it’s as impersonal as house-hold size or age. Modern non-transaction data is what websites you visit, your Facebook friends and likes, the movies you watch, etc. This is the rich personality data that allows for good machine learning and ultimately drives better experiences.
Transactional data is only one part of the story. Without including non-transactional data, companies will not understand their consumers. Transactional data combined with demographics will tell how many of what kind of people purchased a particular item, when. Will it tell you which of those customers have been researching for awhile, which of these customers purchased at this retailer because of a sale, or which customers bought the item because they could not find a better choice, but are not happy about it? These customers have different triggers, access different forms of media, and are influenced differently. According to transactional and even demographic information they fit in the same group. However, that same marketing message will not have a positive impact on this group. Without non-transactional data you can joy understand the “why behind the buy.”
There’s a fundamental problem with transactional data — it’s purely “observed behavior.” And observed behavior is alway missing something — but you never know what.
A good example of this failure is found in paleontology. When I was a boy, dinosaurs were lizards, loners, and scaly. Now we find out many were birds or predecessors of birds, some lives in family groups, and many had feathers. The problem wasn’t a lack of brilliance by the scientists. The problem is they had only observed data — and you can’t know what you aren’t observing. Once they found new fossils, entirely new picture developed.
So transactional data is exceptionally one dimensional and we’ll never know what we’re missing because it’s observed data.
Then the challenge becomes: What does complement transactional data? We must look beyond the big data revolution for that answer. Certainly, there’s some value in other trackable data, but it has the same failing as transactional data (it’s purely observed).
Traditional research increases in importance as a result because both qualitative and quantitative research are needed to fill in the gaps in the observed data. And traditional research is critical to making the observed transactional data make sense.
There is no question that retailers’ use of this non-transactional data is critical for growing personalization needs. This data extends to multiple sources, including social sentiment, news stories, weather impact, local sporting and other events, as well as other external forces, going beyond traditional demographics data.
Using non-transactional data to personalize the shopping experience is still at its infancy, but it is definitely the future of retail. Retailers who understand their customer at the most granular and intimate level will increase the relevance of interactions and improve customer loyalty and increase sales. Customer context is the answer.
BRP defines customer context as “the interrelated factors of customer insights and environmental conditions that make the shopping experience relevant.” Retailers have the ability to know what a customer has in her closet, what she previously purchased, what she browsed on the Web site and abandoned in her online cart, when she is near your store and even exactly what she is browsing and where within the store. In addition to customer insights, customer context considers environmental conditions such as current and forecasted weather, time of day, time of year, media (news), social media, traffic, holidays, events, and other conditions that impact a consumer’s purchase decision.
Retailers must integrate the customer insight and environmental factors with real-time transactions and inventory data to enable and support real-time retail capabilities that leverage “customer context.” To be successful, information must be instantly accessible to store associates. Without real-time data, information provided internally and externally is out-of-date and risks being inaccurate and out of context. Retailers must collect, analyze and respond, in real-time, to their customer’s interactions across all touch points.
With real-time customer context, sales associates can offer customer intimacy and enhanced services like guided selling and personalized offers that are relevant to the individual customer.
NTD dimensionalizes how we see and understand the shopper in many ways far better than past transactions. I would much rather try to connect and engage with the consumer to drive a desired behavior rather than just rely on historical behaviors without context or ability to understand/influence the drivers.
Simple answer is yes. Behavioral analytics are all about taking non-transactional data and using it to predict outcomes and therefore drive messaging to one or to a group. We are just on the cusp of being able to effectively massage the rich stores of non-transactional data being gathered, and the retailer who get there first will have a first mover advantage.
The more sources of data you have, the more complex and accurate analyses and subsequently informed actions you can take. NTD and TD are parts to a whole. I’d say one of the challenges in using NTD is when it is unstructured.