Siebel Order Management Guide > Setting Up Product Recommendations >

Guidelines for Implementing Product Recommendations

The following guidelines improve the performance of product recommendations:

• Avoid heavy eligibility and pricing processing.
  • The product recommendation procedure is executed every time the line item focus changes. It must run very quickly, or it will slow performance.
  • If the customer has slow eligibility or pricing logic (for example, an external call), then configure the underlying procedure so that the slow logic is not executed when called from the product recommendation procedure (for example, by passing an input Mode).
  • Alternatively, remove the eligibility and pricing subprocedure calls from the product recommendation procedure.
• Avoid suppressing duplicate messages for business-to-business accounts.
  • Duplicate suppression processing loads all previous responses for the account at the beginning of the call. Loading more than 100 responses will result in a perceptible delay.
  • For business-to-business selling, configure the upsell and cross-sell message types to suppress duplicates for the quote or order instead of for the account. For more information, see Customizing Duplicate Message Suppression for Product Recommendations.
• Simplify the variable maps used to construct the shared context variables and row set.
  • Product recommendation only needs the Product Id from the current line item.
  • All other fields are used for eligibility and pricing of the recommendations.
  • If you are using a limited version of eligibility or pricing, then create a special variable map that only loads the variables that are required.
• If the product recommendation procedure does not run quickly enough to be called every time a line item is selected (for example, because it calls an external real-time marketing engine), then reconfigure the run-time events that run the engine:
  • Only generate recommendations when a product is picked.
  • Add a button to generate product recommendations on demand.