If the site as a whole is not working well, not getting in the visitors that you want, and not giving the ROI that you're looking for then it needs some work. Sometimes that can be achieved by optimising the site better for search engines but within the existing scheme, sometimes it needs to be rebuilt from the ground up. Without seeing the site in question or the analytics data, it's going to be difficult to get much better than guesswork, but some thoughts:
If the site is performing badly then you haven't got so much to lose. Sure, you might take a short-term hit, but if you can improve the site then that should give you long-term gains, so it will be well worth it. That way, you can address the site structure and page structure, to ensure that it is optimised for search engines and for visitors, and it gives you the opportunity to look critically at all sections and pages to see whether they are pulling their weight or if they could be changed to good effect.
If particular pages are doing badly, why do you think that is? Are they trying to serve a market that doesn't exist or is very small? Are people landing on the pages and leaving straight away? Are they not targeting relevant key words and phrases that searchers are likely to use? Do you actually need so many separate pages or could you serve the business just as well with a smaller, more streamlined site?
The visual design is important to the success of the site, and if you're going to redesign the site structurally, that's a good time to address the visuals as well. But how much is that contributing to the problem? If you have a high bounce rate, it's likely that the visuals are turning people away. But if your bounce rate doesn't seem to be too high then the problem is that people aren't coming to the site in the first place, and that isn't going to be fixed just by improving the visuals.