Categories: Contributor News

Tracking Firefox complaints on Twitter

How do we spot trending support issues even faster?

With the rise of social networking, Twitter has become a common place for Firefox users to bring up issues with their browser. While we have ongoing projects for helping these users, we’ve also been working on a way to use this collection of tweets to let us know earlier if there are new and upcoming Firefox issues. This lets us move faster and get information into the knowledge base and out to our support community as quickly as possible.

Every day, we collect all the tweets that contain the word “Firefox” or “Mozilla” and negative words like “fail”, “sucks”, “hate”. This means we’re looking mainly at complaints rather than overall feedback. We also filter out retweets to make sure we’re looking at users complaining rather than users passing on links. This helps limit the number of tweets to track from thousands a day to a few hundred.

Even with this number of tweets, we don’t have the resources to read them all. What we do instead is to count how many of each word we’re seeing. This is a really crude metric (especially given the sample size), so we can’t use it see what the top Firefox issues are. However. if something suddenly spikes up in number of counts, we can spot it as a trending issue and take quick action.

For example, when Google made the PacMan Google doodle, there was a sharp uptick in the use of the words “sound”, “noise” and “siren” (from about one instance every few days to 15+ in just one day). Combining this sort of data with information from the Live Chat and forum communities lets us spot trends faster and get them in front of developers faster.

These are merely the first crucial steps to incorporating Twitter into our weekly support metrics. As always, we hope to make improvements to the way we present the insights based on feedback and suggestions. Look forward to those changes in upcoming months.

2 comments on “Tracking Firefox complaints on Twitter”

  1. Dave Dash wrote on

    Want me to clusterize this data regularly for you?

  2. Mark wrote on

    I hope you’re including the word “crash” in your negative words.