The CFO of a large industrial company is willing to increase cash outs monitoring due to significant slippage when compared to Operating forecasts. The solution to be developed must include automatic matching of cash outs to initial Purchase Orders and Good Receipts, as well as advanced predictive capabilities. It must also be scalable to include Cost of sales, Inventories and Work In Progress Controlling.


  • 10’s Gb of SaP data, coming from different instances and covering the past 5 years
  • Purchase orders, Good Receipts, Internal matching accounts (401, 408, etc), invoices, etc.
  • A large system migration context with several instances merging during the project


  • Developments of a prototype in Python and PostgreSQL to deep dive in the datasets, with a front end leveraging Microsoft PowerBI
  • Development of predictive algorithms to anticipate cash outs and highlights deviations from what the backlog (PO, GR, invoices) should have generated
  • Workshops and interview with the controlling team to understand and code the specific rules required on internal operations
  • Iterations with the CFO to design the dashboard


  • The dashboard is now available on any devices (including smartphone), with advanced deep dive possibilities: from the cash out on a given BU / commodity to all PO and items
  • The main lesson learnt are related to technological choices and the need to have a clear visibility on data volumes, even when the full datasets are not available at project kick off. Large systems migration is not a show stopper as long as there is clear communication on legacy data continuity requirements