Release Notes Sprint 21 (March 28th 2023)

Here are the new additions with this sprint.

Web app

Here are the improvements made to the webapp

Maximize NBRx as proper objective

To simplify the interface, now the "Maximize NBRx" objective is introduced, instead of having to choose only "NBRx" in the "scope" and maximize net revenue. As a result, the notion of "scope" is eliminated:

All TV response curves now include halo, + tooltip

Any Linear TV tactic that has halo will show the related Response Curve with the halo always included.

Also, the detail of the halo is now provided in the tooltip when mousing over a particular point of the response curve:

Raw data of Response Curves 

It is now possible to download the raw data of the scenario response curves from the scenario page:

With an important benefit: this dataset includes a computation of the initiations

Order or the tactics when editing the constraints

The tactics are now presented in the same order than in the heatmap tables in the tables to edit the constraints, at scenario creation and edition time:

Updating / Duplicating a scenario now always from the same place

Now when editing a scenario the user has to save what has been edited in the pop-in, and then update/duplicate from the bar at the bottom of the screen. This prepares for future UX improvements

Prototyped features

Now it's easier to understand which limits have been reached (if any) for a given tactic when accessing the constraints edition in the advanced settings pop-in:

Ex:

Now in the constraints, both reference and optimized values of spend are indicated. In the max amount field there is the maximum of the response curve as a placeholder, so in this example we can see that the max reached corresponds to the limit of the response curve

In the main KPIs, the absolute amount of variation is now displayed:

Data Science

Improvements regarding the model

The following additions were carried out regarding the model:

• Computation of the halo from AD to NP
• H2 2022 Refresh progress
• Improvements in the output "layout" (dataset used to post back the result of the MMM computation to Databricks)

Improvements regarding the Recommendation Engine (aka Optimizer)

• Halo from AD to NP taken into account
• Computation of ROI for halo (was set to 0 previously)
• Ability to receive the reference budget from the input payload (from the webapp)

Data Automation

• Further progress on the automated integration of HMG data as they were made available
• Continue to integrate weekly queries for the refresh (RGN/SAN L&L/SP, Initiations, Rx