I consulted a few years ago before Yahoo and CNET and my clients were all small businesses, even friends' sites. No matter the size of the project, you can still try to get some insight into your target audiences and what they need or want. I mentioned in a previous comment I used Search once to determine sentiment on a site vs. it's competitors by searching for a feature the site and its competitors all had, along with "like", "love", "hate", "wish", etc. I also took note of who the people were who said those things and where they were talking (forums, twitter, etc). It's a hacked manual approach and although not nearly as quality as a good market research report, at least I have a llittle bit of insight before going out to make site recommendations based solely on tags & links. If you're recommending the site build things that people want (and fix or remove things that they dont), you're more likely to gain links and traffic naturally.
Lastly, it's important to remember that paralysis by over-thinking is a real issue some struggle with. There's no pill for it (yet). Predicting perfection is a fool's errand. Get as close as you can within a reasonable timeframe, and prepare for future iteration. If you're traveling through your plan and determine a soft spot at any time, simply pivot. It's many hours of upfront work to get your strategy built, but it's not too hard to tweak as you go.