What Does It Mean When a Customer Stops Buying Pasta Sauce?: EBM Sparks Up Direct Marketing


Share on LinkedIn

Consider a cash register receipt coupon offer. For many grocers, back-of-the-till coupon generation is either one size fits all or is based upon a simple Boolean “if the customer purchases A, then offer B” model. For example, customers who buy pasta will receive a coupon for pasta sauce. Customers who buy Brand A pasta sauce will receive a discount coupon for another of Brand A’s pasta sauces (or a competing Brand B sauce).

What is still quite rare is for the retailer or sponsoring manufacturer to use past behavior to determine the proper treatment for that customer. If the customer has no history of ever using a back-of-the till coupon or buying pasta sauce, then it is extremely unlikely that the person will do so now.

What if, instead, we actively tracked the pasta buying habits of that customer? Say the customer has a demonstrated behavior of buying five to six pounds of pasta every two weeks and has not purchased pasta in the last 18 days. To bring the person back to the store, we may be better off emailing a coupon (or a double points offer, if the person is not a demonstrated coupon user). Or we could print a coupon for the repeatedly-purchased items that the customer didn’t buy, reacting to the behavior that didn’t occur.

Knowing your customer’s 40th birthday is approaching or that he was recently married can be extremely useful.

Event-based marketing (EBM) is beginning to turn traditional segment-centric direct marketing on its head. The impact is dramatic: EBM programs are typically two to 12 times more effective than traditional direct marketing programs. By monitoring customers’ transactional activity, marketers are now able to micro-target their marketing communications to appeal to immediate needs or react to attrition risk.

While many successful EBM programs are outbound only, the most successful event-based marketing deployments also operate within the real-time inbound marketing paradigm, allowing real-time customer actions to be evaluated within the context of each customer’s long-term behaviors.

Hundreds of behaviors

The critical success factor in a real-time EBM program is the ability to track, in this example, customer-specific grocery cart and coupon-use details. Using purpose-built technologies, such as state engines, EBM practitioners can simultaneously monitor hundreds of customer-specific purchase behaviors (including reward point redemptions) across long time periods while needing to consider only the current day’s purchase activities. In this way, changes to buying habits can be detected immediately.

State engines store observations of only the transaction data, not the data itself. The implication is that the behavior database will only grow as new customers are added. The customer-specific purchase behaviors are refreshed with each visit to the checkout counter.

Data warehouse-based efforts, while logically capable of tracking purchase behaviors, require that the actual purchase detail be stored. Because of the enormous volume of daily purchases, query engines and analytic tools typically cannot sift through the historical purchase data in the available time windows. As a result, warehouse-based purchasing profiles are usually refreshed on a weekly or monthly basis—and the EBM programs suffer a severe drop-off in effectiveness.

But, continuing the grocery example, let’s say we know from our state engine that over the last 12 months, the customer shopped approximately twice per week, is a frequent pasta purchaser, purchases two or fewer bottles of sauce per month, occasionally uses coupons and has redeemed reward points twice. Using this information, the marketer will typically prepare a number of “next best offers” for the customer’s next store visit. In real time, we will use the contents of the market basket to determine which offer is most appropriate.

If the customer purchases pasta as anticipated, we might print a discount coupon offer for the sauce the customer has purchased most—or a directly competing sauce. If the customer does not purchase pasta (and hasn’t purchased any in the last 14 days), we might, instead, encourage the customer to purchase pasta by offering double points. If the customer buys pasta and sauce, we might choose to default to a different offer set entirely. One size never fits all.

Behavioral change is primarily a timing indicator. Contextual life-stage, demographic, profitability and other types of data are always used in conjunction with behavioral change detection as a way to improve business results. In the grocery example, the state engine determines only that a pasta offer needs to be made. Demographic information will typically inform which pasta type (fresh or dried) or which pasta brand (store-brand or name-brand) to offer.

In the financial services industry, knowing your customer’s 40th birthday is approaching or that he was recently married can be extremely useful and important facts to understand when the customer shows a first-time interest in life insurance on your web site.

If you have access to the appropriate transaction data, life events such as a birth, relocation or a job change can often be anticipated with EBM. The birth of a child is often preceded by repeated purchases at baby-oriented retailers or the cessation or reduction of paychecks for the mother. Relocation is often preceded by a change in payroll originator, large check disbursements or high lodging expenses. Depending on the customer’s age, a job change or retirement is commonly indicated by a missing 401k contribution. In certain industries, the actual “event”—such as a receipt of a birth registration or change-of-address letter—is often a de facto confirmation that a sales opportunity has been missed; the customer’s needs have already been satisfied.

EBM programs allow for the rapid and accurate detection of customer behavioral changes relevant to a firm’s business objectives and specific to your customers’ needs. Whether the objective is to drive timely outbound communication or to react in real time to an inbound channel, a well-crafted EBM program is essential to creating an ongoing, proactive, relevant dialogue with your customers.

Dan Smith
Dan manages Practices Communications at Epsilon, and was previously the CMO at both Outsell Corp & ClickSquared. Prior to joining ClickSquared, Dan was the VP of channel development at Unica where he managed Unica's MSP partnerships throughout the Americas. Before its acquisition by Unica, Dan was the CMO at MarketSoft.


  1. Dan, I agree that for too long we’ve segmented customers into categories that may or may not be relevant to a specific consumer. While segmentation may have worked better than “mass marketing,” it’s not nearly as effective as behavioral or event-based personalization.

    Remember the old-fashioned neighborhood grocery store or local butcher—they personally knew you and your family, and could offer “specials” that were most germane to your individual situation.

    Today EBM allows companies to “restore the intimacy,” by making recommendations that are most pertinent to each customer. And that’s a win for everyone.

    Randy Saunders
    Marketing manager for Cincom Systems’ Customer Experience Management products. Randy may be reached at [email protected].

  2. Dan: Your articles contain intriguing ideas, and I see great marketing possibilities! However, the insights you’ve developed from your scenarios depend on the assumption that one retailer can see the whole picture.

    As I read your description, I asked myself, “What would Harris Teeter do?” because where I live in Reston Virginia, we divide our grocery purchases between Harris Teeter, Whole Foods, Trader Joe’s, Safeway, and Giant. When I think about the data that Harris Teeter collects about our family purchases, I believe that they could make some wildly odd conclusions about our food preferences and buying habits. (We cherry-pick brands, products, prices, and other attributes–no pun intended).

    In light of this reality, I think the extraordinary consumer insight you’ve described can result when disparate databases are integrated, and events can be matched. I wrote about this idea in an earlier article on CustomerThink, Your Cutting Edge Strategy Won’t Cut it in 2012, describing the “Universal Identifier,” for consumers. (For example, The UI has the capability of matching Department of Motor Vehicle Records, Real Estate Transactions, Warranty Card Registrations, and other events.) Such matching can provide a more complete picture of the consumer’s lifestyle and current situation.

    Although it seems a long shot right now, do you see the possibility that competing retailers could share information to expand the insight? If not, I see limitations to this approach unless it’s used in markets in which consumers have fewer retail alternatives.



Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here