I come from a brand monitoring world, a space where the potential for many variations of the above can easily happen. Importantly, I'm speaking from an analyst-as-consumer-of-data perspective-- could be a marketer, researcher, client, whomever it is attempting to listen to the data. Point being: ineffective listening is a sin. You should always query and re-query-- as many times as necessary-- so that the data can fully express itself. Data is, in some ways, more responsive than other conversational partners.
As with consumer generated media, when the data universe is vast, participation is unregulated, and the best questions answered are unanticipated, the most frequent sins are probably acts of omission rather than commission. That is, it's not that you go wrong by acting on the data, but in not benefitting from what you could know if you had more effectively listened.
I think clever interrogation is the formula for effective listening: a loop of listening, structuring, iterating and eventually analyzing. Which other effective strategies have you developed to better listen?