Stakeholders Ignoring Model Results

It's always show and tell In initial part of carrer had trouble building trust from stakeholders about ML models, buit a Customer propensity model for target selection real good results in Out of time validation during back testing then senior PM was like if you've serious faith he vouched for our team and challenged the Audience team that just 10% of unique high propensity targets into audience mix for a month would give huge lift and they saw 9% lift the results were shocking to the old age Audience team that did huersitic and then RFM , this and series of no-bullshit conversations helped us there This gave the business trust for us to start building MMM simulator engine, which wouldn't give immediate results because experiments for them can't be run in isolation As a DS or MLE you need to sell your product at times with higher conviction than a typical PM will since if you don't bat for your own creation then who will
So TLDR; Always give proof before asking something in return do your homework run so simulation with back testing data and proceed incrementally (Start small and scale with positive results) for anything since the buisness will never trust a model to be SME