August « 2008 « Blog of Metrics

August, 2008

results — No Comments
21
Aug 08

Explaining a Traffic Spike at mozilla.com

Many within the Mozilla community were excited to see this news about the recent surge in traffic at www.mozilla.com. As great as the recognition is, we’d thought it would be beneficial to take a minute and discuss the exact numbers. In other words, comScore’s numbers are estimates (based largely on panel data), while internally here at Mozilla, we have the actual numbers from our web analytics solution that are likely more accurate and that we’re happy to share publicly.

Below is a breakdown of all the numbers. “comScore” refers to the numbers estimated/published by comScore and “Omniture” refers to the web analytics solution deployed internally here at Mozilla.

As comScore’s chairman points out, comScore’s data is immune from issues related to cookie deletion, so one could expect their numbers to be lower than what companies find through their internal data systems. Still – the discrepancy seems far too big to be described by cookie churn.

So, putting aside the discrepancy discussion, what’s the deal with the June to July spike in traffic?

The June to July traffic increase was driven entirely by Firefox releases (e.g., 2.0.0.15, 2.0.0.16, and 3.0.1). When a release happens, Firefox users automatically see an update page (example here). If we ignore traffic to these pages during June and July, we see that traffic levels to mozilla.com are virtually identical across months.

Uncategorized — 9 Comments
21
Aug 08

A First Look at the Uninstall Survey

The world of Mozilla includes many Firefox related feedback mechanisms. Crash reporting, bugzilla and spreadfirefox.com represent a few such examples. Another feedback mechanism is the Firefox uninstall survey. When a Firefox user (Windows XP only) decides to uninstall their browser, they’re asked if they want to complete a survey during the uninstall process.

Often a user is merely temporarily uninstalling before immediately reinstalling. However, in some cases a user enjoys a seemingly less than optimal experience with Firefox, and in these cases, it behooves us a community to figure out what’s going on and if there are true pain points that the user is experiencing (and perhaps that we’re not aware of).

Since its implementation in March of 2006, well over a million users have visited the survey. Together, they have submitted over 170,000 responses. In a typical day, we might see something in the ballpark of 4,000 visitors, 150 responses, and 40 comments.

Given both the magnitude and value of the resulting information, I’ve been dedicating part of the summer to analyzing and interpreting the data. This blog post is the first in a short series.

So what do the uninstallers have to say? There are two particularly interesting results, the first of which is plotted in the chart below.

After the release of Firefox 2, the percent of users intending to fully replace their previous web browser declined rapidly–from 59 to 31 percent in just one year. The figures for the first month of the Firefox 3 uninstall survey are very similar; 58% of respondents intended Firefox 3 to fully replace their previous browser.

The second result concerns how well each country is represented by visits (not responses) to the uninstall survey. The charts below illustrate the degree to which each of the top eight survey-visiting countries is overrepresented.

Clearly, China is the outlier. Despite comprising only 1.9 percent of active daily users, Chinese users account for 18.6 percent of visits to the uninstall survey. We are still looking for possible explanations as to why Chinese users respond so much more frequently. Are they culturally more inclined to fill out surveys? Do they uninstall Firefox at a higher rate?

The survey data is publicly available at survey.mozilla.com. At this stage, we are looking for suggestions as to how to improve the uninstall survey and, ultimately, improve the user experience.

results — 2 Comments
20
Aug 08

It’s Official – Indonesia Loves Firefox

We’ve previously discussed seasonal fluctuations in Firefox usage, e.g., internet usage typically declines during summer months (June, July and August for North America). Conventional wisdom has held that this behavioral fluctuation is largely attributable to folks in Europe going on holiday and to students in the U.S. and elsewhere spending the summer away from school. Thanks to geoip technology, we’re now able to apply some science to this question and see exactly what’s happening across different global regions.

So, which regions showed the greatest changes in Firefox usage this summer? To answer this, we compared usage levels during a period in May (5.12-5.25) with recent usage levels in August (8.4-8.17). Below are the regions exhibiting the largest changes in Firefox usage:

A few notes:

Firefox usage in Southeast Asia is booming at incredible rates
Contrary to conventional wisdom, the U.S. showed a respectable increase over the summer (3.9%)
Mediterranean areas of Europe seem well represented in the latter numbers

Uncategorized — No Comments
19
Aug 08

Regional Interest in Firefox – within China and India

Google recently released a service called Insights for Search, which appears to be a high powered version of Google Trends. As other keen folks have already pointed out, there’s a bunch of cool analysis that can be done with this new tool – and the display capabilities make the resulting insights potentially much more useful. See here, here and here for some examples.

From Mozilla’s perspective, we can compare the popularity of the search word “firefox” across world regions, time periods, or against other search terms – and with much more drill-down built on top of that. Pulling an initial example of how this might be relevant for Mozilla, I decided to look at “firefox”in two regions – India and China – and see how the proliferation of this search term has grown over time (2005 vs. the past twelve months). Intuitively, such analysis should give us a clue as to which specific areas within China and India have been changing the most in recent years (wrt awareness of Firefox).

India 2005: