Project debrief (post-mortem) - 24.05.

• Speed

• Data maturity

• Simulating different use cases

• Go live plan

• Cha

Attendees: Arthur, Louis

• Too ambitouis in terms of objectve

• We should have been more practmatic, let’s try and then improve from there —> Retrospectively, we did not have much reason to be scared because things ended up well with roty BUT we needed SAP to be good, data clean-up project

• We should have trusted more the algorithm, but we did not have enough information to trust the output → Arthur’s presentation was a game changer

• How can we do light change of data for learning the behaviour, e.g. provide outcome of standard cases (e.g. with different

• Arthur presentation and general documentation
  • Punctual vs. continuous
  • How is the Stock Max compiled
  • Stock Min and how to understand uncertainty

• Better process to pre-assess data format

• Good to see that once the Cajoo data clean-up was done, everything went ok

• Expectations?

• Change of responsibilities at Cajoo was not helping —> We need direct commitment from data/IT resources —> Differentiate connection

• Project management
  • We should have excluded Chilly right away, e.g. Hub Prio
  • Not good approach to know what features to priorities
  • Expectations were not managed not well - there were too many gaps

• Hub Prioritozaton
  • One conversation in Oct
  • Second conversation in Feb that was exactly the same

• Project Mgmt tools → https://linear.app/

• Maybe meetings between the data teams, to avoid conversation

• Big topic: Reporting on customer success criteria, alignment and regular reporting —> Roll out planning, Priorities of product development

• Standard test script was missing

• Internal Cajoo: Never lunch SAP together with replenishment team

• We need to buffers in the project plan

• Data consistency checks througoutg the project