People Change Their Minds Over Time

In this post I will show a few examples of how user sentiment changes over time.  A recent post on the VentureBeat blog describes a company that collects reviews immediately after a purchase is made.  Collecting reviews so near the point-of-sale can be helpful, but the consumer doesn't yet have enough experience with the product to say what's truly good or bad.

For example, consider the Epson Stylus CX5400 printer.  Initially, users were very happy with this printer: one of the reviews was titled "Epson CS5400 Rocks!".  However, after a few months of use this printer started to show its true colors, and owning it went from a joy to a pain in the view of users: the ink jets started clogging up, rendering the printer useless.  Look how the sentiment changed over time in the buzz histogram below:

Electronics are not the only examples where initial views changed over time.  Games are another example, for example, consider Star Wars Galaxies: An Empire Divided. Again early users loved it as "Not just another Hack n' Slash MMORPG".

But, after the manufacturer upgraded the back-end, the sentiment of users changed: "Before the Combat Upgrade this game was really fun for me."

User sentiments for popular web sites also change over time.  In 2004, one user wrote of Neopets.com, "The best site on the web! Better then Nick.com".  By 2006 the sentiment of users changed to "Don't waste your time."  Digging into the reviews, you quickly see that most of the dissatisfaction revolved around the staff. 

While collecting reviews from people soon after a purchase reveals insight into the immediate satisfaction of their decisions, that satisfaction often changes over time. 

This post is really just tip of the iceberg of interesting trends in user sentiment.

Happy summizing.

Comments

[jarmon]

It would be interesting to replace the bars in the buzz display with vertical summize bars and therefore have a "summize-over-time" effect. This would help us see how "polarization" changes over time.

[Abdur]

Sounds like an interesting idea. I will pass it on to Greg, who came up with this visualization of the data. My only concern would be too many colors in a tight space to see the patterns. Keep the ideas coming, we love ideas that help people see new interesting patterns in the data.