Overview (Optimization - Product Recommendation)
- Ondřej Tesař
Optimization - Product Recommendation Accelerator enables you to define the products to recommend to sales representatives – typically when they create a quote and want to have alternatives or offer additional products that make sense in this context. The ultimate goal is to increase quote conversion (by better matching customer needs), increase order size (by offering relevant products) and promote related products (can be manually defined).
Approach
Several types of recommendations are available. The users can configure few parameters and get those recommendations computed:
Frequently Purchased – Bought earlier by the customer and depending on how often a customer bought a specific product.
Others buy – Similar customers (from the same segment) purchasing such products.
Bought together – Products usually bought together, in the same basket or invoice.
Similar products – Products that have similar name, description, or features.
Similar from another brand – Products that are similar but from another brand. The typical use case is to push a specific brand, e.g. private label brands.
Up-sell – Similar products but providing a higher unit margin.
Down-sell – Similar products but with a lower price. The typical use case is to easily provide a cheaper alternative to customers who complain about price, e.g. in a quote revision.
The following recommendation types “Similar products”, “Similar from another brand”, “Up-sell” and “Down-sell” leverage the Product Similarity Accelerator that is required to be deployed. For more information see https://pricefx.atlassian.net/wiki/spaces/ACC/pages/4726882346.
Outputs
The result of this accelerator are Product Recommendations that are stored in tables within the model. Recommendations can be visualized in the last step for a specific scope.
The Product Recommendations can be used anywhere in the Pricefx solution, but specially in Quoting. Here is the out-of-the-box feature to get recommendations:
Or you can configure quoting to add recommendations of a specific product:
Limitations
Product Recommendations are partly based on patterns of past transactions, so it is required to update the model regularly, e.g. once a month, in order to keep up with the latest trends.
Out-of-the-box integration within Quotes provides recommendations for the whole quote. In addition, code examples for recommendations per an item (as in the Demo environment) are available too.
For “Up-sell” and “Down-sell”, assessment of margin and price is not performed for the specific quote context but in average, which can lead to some discrepancies. Additional checking within Quotes might be required.