Scenario Modeling Tip #6: Designate communications to use an agent contact as the channel.
Common misunderstanding: How agent constraints affect general call center agents.
Suppose you want to use a personal banker relationship with the customer. You want each customer to be assigned to a single agent, and you want only the assigned agent to be able to contact the customer for the particular communications that use the agent channel. Therefore, do not use agent constraints for general call center agents (where it does not matter which agent contacts each customer) because it is more efficient to constrain the total usage of the entire call center by using a single constraint. In other words, use agent constraints only when there is a personal relationship between the customer and the agent and no other agent is permitted to contact the customer.
More about agent constraints: Agent constraints are specified in an input data table that specifies the capacity of each agent in each time period. By default, each row of the agent table adds one constraint to the optimization problem. Be sure to specify agent constraints in the Agent input table instead of creating prepopulated constraints in the Constraints input table.
You might be tempted to specify agent constraints in the Constraints table so that you can use communication filters and customer filters beyond what are implicitly used in the Agent table. However, the standard algorithms in SAS Marketing Optimization take advantage of the special structure of typical marketing optimization problems, which usually have a relatively small number of aggregate constraints (for example, at most a few hundred). Specialized techniques are necessary to handle the large number of agent constraints, again taking advantage of the even more specialized structure of agent constraints. The presence of an Agent input table signals to SAS Marketing Optimization that these specialized techniques should be used. It furthermore enforces the special structure of agent constraints that is required by the algorithm, which would not necessarily be the case if the agent constraints were created as prepopulated constraints with arbitrary constraint definitions.
Note: This post was adapted from Michelle Opp’s “Scenario Modeling Tips” document. For more information, see SAS Marketing Optimization User’s Guide.