I’ve worked on a two projects involving design systems (one large for a huge software company, the other for a smaller startup). My main takeaway is that any design system is only as good as the output it creates. (Can I Jab at Salesforce having awesome guidelines and sub-awesome products?). You can have the most amazing toolkit but at the end of the day, if the designers (or developers) consuming them don’t have enough rails or there is too much room for interpretation then you’ve missed your vision.

Other takeaways:

• Plan for it to evolve
• The first version is a starting point (if you try hitting perfection first, you’ll be stuck in review meetings forever)
• Make sure people can contribute effortlessly (don’t get stuck in a world of “backlog of patterns you want to define”, filing tickets, long reviews… V1 and move on)
• Beta test — give your guidelines to designers in your org and ask them to use them to design 3 key UIs each. Then compare the output.