A modern customer management system has two large components: touch points that interact directly with customers and a consolidated customer database that makes information available for decision-making.
The touch-point systems include transaction-intensive call centers and Web sites while the database is often a giant data warehouse with numerous reporting, analysis and modeling tools attached. Connecting these two big pieces is something much smaller: a real-time interaction manager that uses information and strategies from the database to guide interactions at the touch points.
Though it processes fewer transactions than the touch points and holds less data than the corporate warehouse, the interaction manager is still a complicated piece of work. It performs three key functions:
• Linking to the touch points to receive information about transactions and send back its decisions.
• Linking to the customer database (and other corporate systems) to gather relevant background information.
• Building and executing the decision rules themselves.
Each of these functions presents a challenge. Linking to touch points is difficult because each touch-point system has its own data structures and processing flows. It takes work to translate these into a standard format that the interaction manager can process.
Today, this usually involves custom software development, though most interaction management vendors are building prepackaged interfaces for common touch-point products. Linking to customer databases and other corporate resources also requires connecting to a babel of inconsistent systems, with the added difficulty that resolving one touch-point interaction often requires data from many corporate reference systems.
Still, the trickiest part may be building and executing the decision rules. Just how tricky depends on the nature of the rules a given system permits: Some interaction managers limit each rule to one decision at a time, while others allow rules that execute a sequence of events. Some systems explicitly predefine each path, while others let multiple rules interact to choose the best response. Some rules scan for changes in behavior patterns, some adjust automatically to user reactions, some dynamically assemble custom replies.
An ideal system would provide all these capabilities as well as empowering nontechnical users, simplifying control over rule deployment and deactivation, linking seamlessly with touch-point-specific content and providing real-time information on results.
Sadly, this ideal system does not yet exist. Each of today’s interaction managers meets a different subset of these requirements. So it is up to each buyer to decide which capabilities are most important and find the system with the closest match.
PathWay Solution (YellowBrick Solutions, 919/653-2300, www.yellowbricksolutions.com) provides one intriguing mix of interaction management capabilities. PathWay is a new system, released this month, but is descended from the pioneering Persimmon IT TargIT interaction manager released in September 1997 (and reviewed here in December 1997).
The heart of PathWay is what the vendor calls Visitant technology, which stores pointers to information about each customer in one or more underlying databases. In addition to the usual data, such as demographics and transaction summaries, this information includes the customer’s current location in whatever multistep marketing campaigns it belongs to.
Calls to PathWay are embedded in touch-point systems at specific locations, such as a slot on a Web page; when a customer reaches one of these locations, the touch point sends the PathWay the customer ID, a list of campaigns associated with that location and perhaps other information such as replies to survey questions. PathWay then finds the customer’s Visitant, loads its data into memory, determines which messages are due in the specified campaigns, finds the related content – say, an HTML snippet for a Web page – in a content repository and returns the content to the touch point for delivery. All this, of course, happens in the blink of an eye. The system then keeps the Visitant in active memory for a reasonable period so it need not reread the database if there is another transaction from the same customer immediately after.
Though Visitant technology lets PathWay operate across different touch points, the campaigns themselves are what give marketers precise control over how each customer is treated. Each touch-point location can call for one or more campaigns, and each customer can be assigned to one or more campaigns. This combination of choices offers the marketer tremendous flexibility.
If all locations call the same campaign, then the customer will see the same message regardless of how he connects with the company – providing much-desired cross-touch-point consistency. If each location calls a different campaign, then the message can be tailored to the specific situation. And if different customers are assigned to different campaigns, then the same touch point can give each customer a tailored experience. Furthermore, even a single campaign can include branches to deliver different messages based on customer data, current behavior and other conditions. And since the Visitant keeps track of the customer’s location within each campaign path, PathWay can deliver a coherent sequence of messages across a disconnected set of contacts.
Naturally, PathWay includes a campaign design tool to construct these campaigns. This tool, Experience Manager, is reasonably easy to use, although the process of linking campaign steps to touch-point-specific contents is rather cumbersome. Customers are assigned to campaigns using a segmentation tool, called Profiler, which also provides basic reporting. The vendor is working to allow use of other segmentation and campaign definition tools with its Visitants, although this hasn’t happened yet. It also plans to incorporate automated statistical modeling capabilities; at present, only precalculated model scores stored in the underlying customer database could be used within a campaign. Dynamic scoring would help PathWay address one of its current major weaknesses: the inability to return the best campaign when a customer qualifies for more than one. Currently, all campaigns at a given touch-point location are ranked in a fixed priority scheme that is applied to all customers. Some method to find the best campaign for each individual customer is highly preferable.
PathWay is priced at about $150,000 for the full suite of components. It runs on Windows NT and Sun servers and uses the SQL Server and Oracle databases. Initial client installations are under way.
