David Raab Article Archive

The last few articles have discussed the “interaction manager”–a central system that coordinates contacts with customers across independent touchpoints. The key role of the interaction manager is to make decisions about how to respond to a specific situation. But these decisions must be translated into specific content before they can be presented to a customer. Handling this efficiently turns out to be one of major challenges in the interaction management process.

The form of the content depends on the touchpoint system–it may be an HTML page, telemarketing script, direct mail letter, WAP message, and so on. Content for each touchpoint is usually created and managed independently, either in the touchpoint system itself or in a specialized system such as a Web page development tool. The interaction manager itself rarely performs this function, although a few systems do manage email messages internally for user convenience. Content management itself is a demanding operational process with extensive requirements for text and graphic manipulation, version control, authorizations, work flow, controlled deployment, staging, expiration dates, geographic and other distribution constraints, security, and other functions.

But while the interaction manager does not actually manage content, it must somehow ensure that its decisions can be delivered through whatever content is available. At one extreme, the interaction management rules may actually include references to specific pieces of content. This requires that the user know what content is available when the rule is being defined. The communication may be handled manually (by having the user look at the content management system), automatically (by having the interaction manager read the list of available content items from the content manager), or in some hybrid fashion (perhaps by having the user manually enter the list of available contents, or having the system import the list periodically). The problem with any of these methods is that the rule must be changed if the appropriate content changes–perhaps because the old content becomes obsolete or unavailable. If the content reference is physically embedded in the rule, this reference may be hard to find and in any event must be manually modified. At best, this is a lot of work; at worst, the system will ask for content that is unavailable or inappropriate. This approach also makes it difficult to add a new touchpoint, since every rule must be modified to add the new content references.

The other extreme for the relationship between content and interaction management is for the interaction manager to pass back a generic decision, and let the touchpoint (or content manager) determine what content is appropriate. This shifts the translation burden away from the interaction manager, although the fundamental problem still remains of how to determine which content relates to which decision. In some ways, moving the responsibility to the touchpoints makes the task even more challenging, since touchpoints are less centrally managed than the interaction manager. This means that when a new decision–say, a new offer–is added to the list of possible outcomes that can be returned by the interaction manager, each touchpoint must be updated separately to execute that decision. But this approach does simplify the problem in other ways, since the touchpoint system is inherently aware of changes in available content and can make adjustments without the interaction manager getting involved.

Other approaches are also possible. In some arrangements, the interaction manager passes back the actual message to be displayed–say, a bit of text with a product name–which the touchpoint then puts into an appropriate format and delivers. This works well enough but is limited to fairly simple communications. Another approach is to define a standard set of attributes to describe content, and have the interaction manager return attribute values instead of references to specific items. The touchpoint or its content manager can then find whichever available content comes closest to meeting the specifications. This approach avoids the need to explicitly link a particular interaction management decision to a particular piece of output, but does require that all touchpoints employ a consistent set of attribute definitions. It also requires a mechanism to choose similar content when no exact match for the specified attributes is available.

A more radical solution is to create a central content store that holds content for different touchpoints, and perhaps even to create multi-media documents that hold different forms of the same offer. This is possible with object technologies or XML documents. Mapping interaction manager outcomes to a single content store is considerably easier than mapping outcomes to each touchpoint individually. Still, methods that avoid explicit mapping altogether will ultimately be needed to avoid an administrative burden that will stunt large-scale deployment.

All of these approaches are used today in some degree, although the most implementations still rely on explicit content specification within rules or custom integration between the interaction manager and touchpoint. Well designed systems also provide fail-safe mechanisms that display default contents if no valid reference is received. They also report on invalid references that allow administrators to research and correct such conditions after the fact. Systems may further help administrators by providing lists of all rules that employ a given piece of content and allowing universal search and replacement of content references. They may also let the interaction management administrator view available contents to ensure they include the intended messages.

Some interaction managers also provide functions related to multiple offers in general and to Web banner ads in particular. These decision rules must consider not only the current customer’s interests, but also the number of items that can be displayed simultaneously; methods to rank, sort or prioritize the items; the number of impressions, click-through or unique viewers that have been promised and already delivered; the number of times a customer is allowed to see the same item and the minimum interval before an item can be repeated; the revenue associated with different items; and so on.

* * *

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.