In this tab, you will find a presentation of the architecture of the back-end. This includes both the inputs and outputs used by the back-end.
Main data sources :
SAP BW :
Transactional data updated monthly. This is the source used to define the scope of CPC entering the AI engine. It is also used to gather the current prices of CPC and to compute several features.
SalesForce :
Main source of data for customer and products related data. Used to compute features in the AI engine and display dimensions in the front-end tools.
AI smart price engine :
Also called "back-end tool", the AI engine is built using DataIku. This is where the price recommendations are being computed. Please refer to the documentations (business presentation here) for more explanation on how the AI engine is recommending prices on a CPC level.
The output of the model is stored in a BigQuery project.
Front-end tools :
Two front-end tools are available. Most of the analysis are expected to start from the dashboard that allows to drill-down on subsets of CPC. WebApp is then used to understand the recommendations at a CPC level.
Selective Pricing Optimization Tool dashboard :
Qliksense dashboard containing the general analysis of the recommendations of the AI engine. The user guide of the dashboard is available here.
The dashboard is plugged directly on the BigQuery project containing the results of the AI engine. It also contains direct links to the WebApp.
SPOT WebApp :
The WebApp answers the need of more detailed analysis on a CPC level. For every CPC, it displays the comparable set used to generate the price recommendations. This allows the end-users to understand the logic behind the recommendations and decide if they are accurate or not on a CPC basis.
When a comparable set is considered inaccurate, the users are able to deselect comparable CPC and have the recommendation being updated.
The WebApp also allows admins to extract the whole dataset behind to do some consolidations. In the campaign process, this extract can be used to generate a file that will be sent in SalesForce campaign module to initialyze a campaign.
As the dashboard, the WebApp is plugged directly on the BigQuery project containing the results of the AI engine.
Expected process for a new campaign :
- Define the scope of the campaign and the variables to use (date scope, original price used, etc.).
- Run the AI engine in DataIku
- Validation of the outcome of the run (model performance, price impact, number of cpc, etc.)
- WebApp and dashboard synchronization (dashboard will be updated daily, WebApp has to be synced manually).
- Business reviews using front-end tools, comparables deselections in WebApp and consolidations using WebApp extract.