There aren't really any "guidelines" but there's a lot of information floating around about what designs are good for conversion. You can learn a lot from abtests.com, which often talks about changes made to a product homepage that have led to an increase in sales.
Best practices I've come across and applied to the design of my product's promotional site:
Know your audience. You can learn from design patterns and conventions but no one knows your audience like you do. Many products, like productivity suites, have a very broad audience of non-technical people. My product is targeted exclusively at technical people. The same lessons for the former category need not necessarily apply to the latter. Always make sure you're aware of why you do something and how your audience might react.
Watching a video is a long term commitment. Look at it like this: your product homepage is competing for my attention with every other website on the Internet and anything else I might be doing. So asking for 2 minutes of my time is a big commitment. If you put video on your site, it has to be as a fallback - I should be able to see the exact same information in static text and images so that I can glance at them at my leisure.
Don't be anonymous. Instill confidence by putting brand logos and customer testimonials on the page so I know I'm not the only person signing up for this. Social proof is a very powerful persuasive element (which is why "A+, would buy again" is so effective).
Put a clear call to action in view above the fold. 37signals has written a lot about their tests with different calls to action and they ended up going with "see plans and pricing" because it set expectations about what would happen when you click, as opposed to "try free" which raised the question of whether you'd have any commitments, such as having to fill in credit card info. Additionally, leading prospective customers through the pricing page increases conversion to paid accounts.
Carefully consider copy. Mentioning 37signals again, they once wrote a blog post about headline changes for their Highrise product. The headline that set expectations and used an imperative tone had the biggest improvement in conversion. Key takeaway? Set expectations as explicitly as possible and tell people what to do (or take away doubt, depends on how you look at it).
Further, once you launch your design, it's important to continue iterating and testing. A/B tests are an excellent way to do this. A product like Performable will also take you another step of the way and help you figure out how to track the "customer lifecycle", eg. from the moment someone arrives on your site, what do they do and when do they leave? And you can try something like the five second test to get some quick remote usability feedback (great weapon in an argument, and by the way, doing some five second tests on that site yourself can be quite educational as you'll see a wide spectrum of alternative designed-for-conversion marketing sites).
Finally, some food for thought: 37signals (again) is testing new designs for its Highrise page that includes no images and puts the call to action at the bottom of the page which you can only see after scrolling 5-6 pages down. Since these are currently being multivariate tested, there's no proof that these have greater conversion than those with calls to action and images/video at the top of the page. But it makes you wonder, right?