Results in the company
I'm starting a recommendation project for a retail client, and it's the usual conflict: the business wants to see results in a month, but I know that without a sensible data pipeline I won't build anything stable in that time. How do you handle this so you don't end up with a model built on garbage data just to show a slide in a meeting?
1 comment
ju4417
·1 month agoThis conflict is probably inevitable, and the only thing that helps is showing something quickly, even if it's simplified, to build trust, and separately explaining that this is not the final version but a starting point. I usually build a simple model on the available data in the first week, show the direction, and then over the following months improve the foundation. You just have to clearly communicate that this first version will be limited. It's well described here in the context of combining a technical approach with business expectations: https://zlotowskie.pl/artykul/synergia-algorytmu-i-n1803969