That's when you need to look back, and not just focus on the most popular complained-about things, but also look at the qualitative feedback you're getting, and they to synthesize commonalities between different lines of feedback. Like another commenter said, sometimes feedback will help you reach a local maximum, but you need to infer /why/ complaints come in the first place, now just what people are complaining about, to know where to go next. If you have so many complaints and there's no clear direction, either you're at the nitpicking phase, or there are larger problems that you haven't identified yet.

If your current tagging system isn't helping you, flip the axis on which you're tagging to see if new patterns emerge. (area of complain on your product, versus types of tasks users are trying to achieve, for example.)

We've been using User Voice for the computer game Torment: Tides of Numenera. First, we used it to get community feedback on our rewards, tiers, and approach for our Kickstarter campaign. It was very helpful in testing our ideas in advance.

Throughout development, we've been using it to get feedback from the community: on their own ideas, their thoughts on design aspects we've shared, and questions they'd like us to answer.

User Voice lets people suggest and vote on ideas. We can label them with customizable tags like "Seen," "Considering," "Not Planned," as a very time-efficient means of communicating to many at once, without the expectation of personal replies.

The system isn't perfect, and we're not using it perfectly, but it's been helpful for getting and organizing feedback.

That's why I really liked Dell's Idea Storm. Instead of worrying about what a single user thinks, you can only worry about the current most popular issues or requests.

I don't think it has really helped Dell as to many of the things they simply said they aren't doing it (e.g. standardise power supplies in laptops) but the concept seems decent if they weren't very open to the feedback received.