AI Assistant (Leat MCP)
July 20
→
Margin-aware decisioning
Compare each action’s expected result with its full cost, then select the option likely to contribute the most after that cost is deducted. Redemption and response rates still inform the decision, but they are not treated as the goal. A widely used incentive is only effective when it creates enough additional value to justify what the business gives away.
Each candidate action carries the costs relevant to the customer and transaction. This can include the face value of a voucher, the discount applied to the current basket, the future cost of points issued, the product cost of a free item, revenue the reward may replace, and the cost of the delivery channel. These costs become part of the score rather than being reviewed after the decision.
For each candidate, Leat estimates the additional result it could produce, the margin associated with that result, and the cost of the action. The remaining value becomes its expected contribution. A €5 voucher can therefore rank above a €10 voucher even when the larger amount has a slightly higher response rate, provided the smaller incentive is expected to leave more value after cost.
Define the values available for an incentive, such as €3, €5, and €10, then let the Decision Engine choose the smallest amount expected to create the desired change. It can also compare different types of value. A high-margin free item may rank above a discount, while a no-cost message may beat both when it is expected to work just as well.
Margin Protection sets the firm limits, including the minimum margin a transaction must retain and the maximum cost an incentive may create. Margin-Aware Decisioning ranks only the actions that remain within those limits. Each decision records its expected contribution, while Optimization compares expectations with actual transaction results and controlled tests across decisions, customer groups, and incentive types.
Incentive cost modeling
Points, vouchers, products, discounts, and delivery channels can be assigned costs so the engine compares them on the same basis.
Cost data maintenance
Product costs can remain current through POS and e-commerce data, with missing values completed at product or category level.
Incentive value ladder design
Teams can define several available values so the engine can choose the smallest amount expected to achieve the desired result.
Reward substitution review
Teams can review when free items, messages, or other actions rank above discounts and confirm that the expected margin supports the choice.
Margin contribution reporting
Expected contribution can be compared with measured results across decisions, customer groups, and incentive types for finance and program reviews.
Margin-based campaign testing
Response-based targeting can be compared with margin-aware decisioning through a controlled test using cost per additional order and contribution as measures.
