Just like hitting a baseball, timing is everything when talking to your customer. If you don't provide the right message using the right delivery vehicle at the right time, chances are your customer is going to ignore your hard-crafted pitch. In an article in the November issue of Harvard Business Review, authors Kirthi Kalyanam and Monte Zweben envision a computer-based model called "dialogue marketing" that interactively tracks and communicates with customers at the exact moments the customer deals with the company, be it during an address change or the purchase of a baby seat. "Its principle advantage over those older approaches," the authors write, " is that it is completely interactive, exploits many communications channels, and is 'relationship aware'. . . ."
This excerpt discusses how a company might launch dialogues during customer "transitions," such as when they trigger a loyalty program or are poised to defect to a competitor.
The nature of changes
No two companies collect identical data, and so each will define the specific transitions that trigger customer dialogues. In general, transitions reflect a change either in the customer's life or in his relationship with the business. They are particularly valuable in the creation and operation of complex loyalty programs. A company might set dialogues to launch when a customer:
- makes a purchase in a new category (first suit, first time in our new shoe department);
- buys a brand so frequently that he qualifies as a zealot;
- exceeds a certain monetary value in a year and thus qualifies as a VIP;
- crosses a threshold in a loyalty program;
- requests a change of address, indicating a household move;
- purchases an item that indicates a major life change such as an infant car seat, or a large project such as a sink or cabinets. (Dialogue systems can also incorporate data from third-party vendors that track the issuance of new mortgages, openings of college savings accounts, and other commercially available data.)
Once a transition triggers a dialogue, that dialogue often moves the customer to another transition, at which point a new dialogue with a different goal kicks in. For example, a major regional grocery chain identified a tipping point: People who ordered from its Web site more than four times were very likely to become regular customers. The company's technology staff programmed its dialogue-marketing system to respond to new customers via e-mail, using steep store-financed discounts to push them aggressively through those first four purchases. After purchase number four, the system automatically begins generating less expensive treatments, for example, offering only trade promotions financed by consumer goods manufacturers.
| Customer value is a major consideration in channel choice. |
Other transitions represent not an opportunity but rather a threat—usually, that a customer will defect. Recall for a moment the airline with its high-value customer gone AWOL. Now imagine the company practices dialogue marketing. The system detects the customer's failure to book a flight for several months and sends a message to a service representative asking him to call and try to rectify the problem. If the customer complains of poor service, the representative logs the details into the system, triggering the creation of a written apology signed by an executive and including an incentive to return-automatic upgrades or frequent-flier miles, for example. If her account remains inactive, then the next month the system sends her another note with a reminder about the incentive. If yet another month passes, she gets a call from a senior customer-service manager. When she finally buys a ticket, the system launches a dialogue for special treatment of "saved" customers.
As this example illustrates, dialogue marketing considers not just when to communicate but also how to communicate: Dialogues play out across multiple channels depending on the content of the message and the urgency of the situation. Some customer actions will trigger the system to respond immediately, with a personalized e-mail, a personalized piece of direct mail, or an alert to a salesperson to make a call. Other responses will lay the groundwork for future interactions. The system might store a personalized message at the point of sale to be delivered when the customer next visits, or send the message to a call center representative to be used when the customer phones. Or it might place customer-specific content on a Web site in preparation for a virtual visit.
Customer value is another major consideration in channel choice. Before launching its resource-intensive retention dialogue, the airline's system would establish that the customer in question is top tier, based on her previous and predicted purchasing. If the customer qualified as only medium value, the system might mail her a message, either electronically or by placing her contact information on a direct mail list together with personalized content for the message. A low-value customer might receive an e-mail, or nothing at all if he is not profitable or not online. This portfolio of approaches is another way in which dialogue marketing improves on most one-to-one initiatives.
The data rich and the data poor
Not surprisingly, companies that collect masses of data and cut that information every which way have emerged as dialogue marketing's virtuoso early adopters. Harrah's Casinos, for example, which is known for its sophisticated use of database-marketing technology, runs an elaborate set of dialogues that consider every permutation of the casino-vacationer relationship (see Gary Loveman's article "Diamonds in the Data Mine," Harvard Business Review May 2003). Harrah's system defines customers in these ways:
- Decliners: number and recency of visits are dropping
- Incliners: number and recency of visits are rising
- Inactive: no visits for a long period of time
- New Business: only one recent visit
- Past Due: no visits for a designated period of time, but not yet considered decliners
- Due Now: not yet considered decliners but need a tickler to bring them in now
- Ad Hoc: everyone else
The company uses dialogues to take customers from negative states (such as decliners) to positive states (such as incliners), chiefly by alerting live "player hosts" to intervene at the first sign of a transition. Since Harrah's began using dialogue marketing in 2003, it has raised revenue by 30 percent to 69 percent, depending on the customer segment; increased by 20 percent to 25 percent the contacts that produce return visits; and boosted the number of VIP customers overall. 
Technology: What Lies Beneath
Four fairly new technologies make dialogue marketing possible:
Intelligent Process Engines. Standard database applications store multiple tables of data and provide screens that allow users to insert, edit, and delete entries from those tables. Intelligent process engines do more: They track individual "states"—where a customer stands in relation to the company at any given moment. (For example, a person who accepts a promotion is in a different state than a person who rejects it.) These engines initiate actions based on changes to those states and provide feedback to managers on the results.
Event-driven Computing. Intelligent process engines respond to changes in states; event-driven computing responds to actions as they are noted by a company's enterprise systems. An automated engine runs continuously in the background of all customer-focused applications, "listening" for events (a customer playing a slot machine; someone registering on the Web site) and deciding how to react, thus minimizing the need for human intervention.
Scalable Web Application Architectures. Dialogue marketing requires companies to manage dialogues for millions of customers who are in different states of relationship with the business and to do so across multiple channels. Scalable Web application architectures are distributed systems that balance the load of user requests across many computers. As computational demand increases, these architectures allow companies to add more servers so they can incrementally scale applications to serve those millions of customers in a variety of ways.
Web Services. In dialogue marketing, multiple databases and other systems must constantly talk to one another, a task that once required significant programming. Web services allow one computer to send a request for information over the Internet using a simple URL address and another computer to automatically respond in a standard language called XML. This advance greatly reduces the amount of computer skills required to operate such systems.