How much modeling do experienced HF analysts/PMs actually do?

I'm curious how experienced hedge fund analysts and PMs think about financial modeling over time.

When you're evaluating a new idea, do you still spend a lot of time building detailed models from scratch, or do you mostly use the sell-side model as a starting point and focus your effort on changing the key assumptions, scenario analysis, and sensitivities?

As someone earlier in my career, I've found that building a clean model from scratch can be a huge time sink. While it definitely helps with understanding the business, I'm wondering whether the ROI declines as you gain experience.

For those with several years on the buy side:

• Do you still build models from scratch regularly?
• In what situations is it worth the time versus simply auditing and adapting the street model?
• Has your process evolved as you've become more experienced?
• If you inherited a messy or unreliable sell-side model, would you rebuild it, or only rebuild the sections that matter to your thesis?

Would appreciate hearing how analysts and PMs at fundamental L/S funds actually approach this in practice.