Incentive Recommendation Metrics

Contents

• The Incentive Recommendation dashboard
• How do we calculate the numbers on the dashboard

The Incentive Recommendation dashboard serves as a monitoring screen for the feature.

The Dashboard is divided into four sections:

• Campaign revenue - Displays the revenue attributed to campaigns using Incentive recommendation, on a daily break-down.
• Estimated savings on incentives - Summarizes the savings achieved by using incentives.
• Lifecycle overview - Displays the overall purchases made by customers in different lifecycles.
• Engagement - Displays a breakdown of messages, opens, clicks and purchases.

The first step is to define when is a purchase attributed to an Incentive Recommendation campaign.

The method we use is similar to to what Predict uses. First we examine whether there were any clicks within a 7-day period before the purchase. If not, then the purchase is not attributed to any campaign.

If there were clicks, the closest one to the purchase is considered and it is attributed to that campaign. After all attributing has been done, we select the ones that used Incentive Recommendation.

In the Campaign Revenue widget, the following data is displayed:

• All revenue - This is the sum of all purchases attributed to the incentive campaign launched on a given day.
• Smart incentive revenue - If measurement is turned on, this is the sum of all revenue coming from contacts who received incentives distributed by the algorithm. If measurement is off, this is the same number as All revenue. For more information on measuring Incentive Recommendation campaigns, see Measuring the effectiveness of smart incentives.
• Control group revenue - If measurement is turned on, this is the sum of all revenue from contacts receiving the benchmark (control group) incentive set by the customer.
• Messages sent -  The sum of all messages sent.

In the Estimated Saving On Incentives widget the following data is displayed:

• Savings - We calculate this at contact level. When a purchase occurs, we take the difference between the highest possible incentive in the campaign and the actual purchase price. This difference is the saving. All savings for a given day are summed up and displayed.
• Messages sent - The number of all messages sent.

Here are two examples:

In this case we presume that the incentive was used. We calculate a theoretical full sales amount, that is, the sales amount plus the discount the customer received from the incentive. Using this we calculate the discounted price with all incentives that the customer could have used (having a minimum cart value less than the theoretical full sales amount).

The saving is the difference between the actual sales amount and the minimum of the theoretical discounted sales amounts.

Example: The customer gets a $10 discount if purchasing above $75. The other two incentives in the campaign are a discount of $5 above $50, and $20 above $100.

Let's say we see that the customer makes a purchase with a sales amount of $95. We assume this is the discounted price. The theoretical full sales amount in this case is $105.

This is above the $75 limit so we assume that the customer used the incentive. If they had received either of the other incentives they could also have used them, since their minimum cart values are below the theoretical full sales amount.

The discounted theoretical prices are: $100 for the $5 incentive, and $85 for the $20 incentive. Thus, the saving is the actual discounted price ($95) minus the maximum discount price ($85), giving us a saving of $10.

In this case we presume that the incentive was not used. Thus, the sales amount is undiscounted and it equals the theoretical full sales amount. From this point the calculation is the same as in the first case. In practice this means that the incentive in this case acts as a trigger for purchase without costing anything, just like a zero incentive.

Example: The customer gets the same discount as above ($10 discount if purchasing above $75). The sales amount was $60, which would have given a theoretical full sales amount of $70 had they used the voucher, except that this is below the $75 threshold, so we can be sure that the incentive was not used. The theoretical full sales amount is then $60.

There is one incentive that they could have used: $5 above $50. Had they received this incentive the sales amount would have been $55. The saving is again the actual (not discounted) price ($60) minus the maximum discount price ($55), giving us a saving of $5.

In the Lifecycle Overview widget the following data is displayed:

Here we summarize all purchases of customers, based on their current lifecycle. For more information, see Customer Lifecycle Report.

In the Engagement widget the following data is displayed:

• Message deliveries - This number reflects all messages delivered to contacts' inboxes.
• Opens - This is the number of all message opens.
• Clicks - This is the number of all clicks in the messages.
• Purchases - This is the number of all purchases attributed to the clicks.