On his blog, Steve Wexler posted a nice overview of different ways to present Likert scale survey data. Go take a moment to read it and come back.
Back already? OK.
The one thing I have to add to Steve's overview is that in most cases the statistic to focus on is the top-box score, the percentage of people who gave the highest possible answer such as "Very Satisfied" or "5" out of five.
In every case we've looked at, getting customers into that top box is what drives business goals. For example, in the graph below (which I shamelessly stole from our recent NCSS Mobile Phone report), we show that in the mobile phone industry, it's really only the "Very Satisfied" customers who are loyal. 79% of the top-box customers say they would subscribe again, but that number drops to 11% for the "Somewhat Satisfied" customers.
So for purposes of driving customer loyalty, the "Somewhat Satisfied" customers are not that much better than the "Very Dissatisfied" customers.
And while this particular graph is for the mobile phone industry, we see the same trends over and over again with every survey we run.
It seems that for the vast majority of customers, if they don't give you the best possible score it's because they have a reason. It might be a small reason, but that small reason is enough to make them interested in looking elsewhere for their business.
(As an aside, it's excellent practice to follow up any less-than-top-box answer with a question about what the company could do to make the customer "Very Satisfied.")
So when you decide how to present survey data remember that the reader's attention should be drawn to the top-box score, since that's the most likely to impact the business goals.