I agree. But startups shouldn't be expected to do this. Startups are experiments; you want to quickly determine if there's a market for the product (and fail fast if not, so you can alter the product or move onto something new). Being conservative with design is the perfect approach for this. If you experiment with unproven layouts, and it fails, how do you know if the problem was the product or your layout?

Interactive new stories, artists' portfolios, design agency websites, web apps, and websites that are themselves art - these are all good places to push boundaries in design. Not startup product promo pages.