Bottoms Up!

User insights are an essential part of creating successful marketing programs that hit the mark.  As a result of a recently concluded research project, I believe we are well positioned heading into 2009 with a better understanding of who our users are – and aren’t– and what we can do to continue to grow the base in the new year.

Getting to this point began with a simple question: if Firefox has 20% worldwide market share, what about the other 80%?
•    Who are they? How are they different than our current, loyal user base?
•    Why haven’t they downloaded Firefox yet (or have they)?
•    Have they even heard of Firefox – if so, what do they think about it?
•    What types of activities and tasks do they perform online?
•    Where and how might we connect with them to introduce Firefox?

To get at these questions and more, we recently conducted an Internet market segmentation study. By better understanding the Firefox user base within the broader context of the overall Internet user population, we hope to discover what makes the “other 80%” different from Firefox users. And in doing so, formulate some new ideas for continued growth.

Background [if methodology makes you yawn, skip down to Key Learnings]
The objective of any segmentation research project is to identify distinct sub-groups that exhibit different behaviors, attitudes and perceptions than other segments. In the case of the web browser market, we essentially wanted to take a snapshot of the entire Internet population at a moment in time to see what segments exist and what we could learn from them.

Unfortunately, it is not possible to do segmentation on a global scale – the estimated 1.4 billion Internet users worldwide do not lend themselves to a research methodology at that scale. Even in large markets like the United States, the total market size presents a challenge without doing some degree of sampling. If you know anything about how long it takes the government to collect census data, you’ll appreciate that we took a few shortcuts.

The first was to limit scope to the US. No slight to any other region intended – in many non-US markets Firefox share is much higher – but the objective was to execute this phase of the research without introducing the complexity and cost to replicate on a global scale. We worked with comScore, a well-known market research vendor with a consumer online “panel” of one million users in the US, large enough to serve as a representative sample of the market.

A few bits to note:
•    panel data is a sample; just like all the political polls leading up to election day, there is a +/- margin or error to consider;
•    the comScore panel is opt-in – consenting participants are aware their online behavior is being tracked for research purposes;
•    the panel is made up of Windows PC users only – the Mac and Linux population is accounted for as an overall % of population but their actual online behavior is not captured;
•    for more about comScore’s panel, visit comScore methodology.

The decision to use a 3rd party panel like comScore was driven by practicality and timeliness – their primary business is collecting behavioral data, normalizing and extrapolating it to represent the online activity of the broader population. For this project, comScore created a US segmentation matrix that crosses US Firefox users against all US browser users. The distinct segments are categorized by time spent (total minutes) as a proxy for browser usage intensity and engagement. The cells are defined as:
•    Super-Heavy = top 5% of users
•    Heavy = next 15% of users
•    Heavy-Moderate = next 15% users
•    Light-Moderate = next 15% of users
•    Light = bottom 50%

The resulting matrix is a way to cluster “like” groups together to better understand their population size, online usage patterns and most importantly what sets them apart from other segments. The first phase of the process began in September looking at July 2008, the first full month of data following the release of Firefox 3.0 in mid-June. The second phase concluded with an in-depth survey of the three primary segments identified.

Whew — that’s a lot to say, but it’s important. What follows are some of the highlights that might be most interesting to the Mozilla community. Some findings are consistent with data we’ve collected from other sources, such as NetApplications, while other insights are new and need to be further vetted and scrutinized. But the overarching point is that even directional insights are better than none at all (so the theory goes).  So without further delay, here’s a bit of what we learned.

Key Learnings, Part 1
For the month of July 2008, there were approximately 29.3 Million US Firefox users representing roughly 16% of the total US Internet population. For the month, comScore estimated the total US population at 189 Million, which is notably higher than Nielsen Netratings July total of 165 Million, but largely in line with eMarketer’s comparative roll-up of 193 Million users.

Interestingly, this estimate of total unique users translates into a disproportionately higher percentage of monthly browser usage:
•    24% of total Page Views on Firefox (avg. of 3,650 PVs p/user)
•    22% of total Minutes on Firefox (avg. 2,283 minutes p/user)

Given that we typically think of Firefox users as more tech-savvy and online intensive, it is not such a stretch to believe that both engagement metrics – Page Views and Time Spent – are above 20% and thus, in line with US market share as measured by NetApplications. What gets somewhat more interesting is the break down of total population into discreet segments.

Usage Segments
At first glance, the segmentation matrix (chart #1 below), looks like a 5 x 6 patchwork quilt. The horizontal axis represents the total Internet population, while the vertical axis represents US Firefox audience. Below the line is that “other 80%” of the US Internet base with no Firefox usage in the monthly snapshot.

comScore US Internet Segmentation matrix, July 2008

For each cell in the matrix, we collected the following data:
•    Total unique users (UUs)
•    Avg. Usage days
•    Avg. Page Views
•    Avg. Minutes
•    Avg. Browser sessions

The color-coding shows how we’ve clustered similar segment cells based on browser type (Firefox vs. non-Firefox) and usage intensity (online minutes). Then, we named the discrete segments for the strategy that might be employed with that particular segment. Here’s how:
•    Green = Retention segment using Firefox for majority of online minutes;
•    Blue = Engagement segment uses Firefox but not for majority of minutes;
•    Orange = Acquisition segment with no Firefox usage in a given month;
•    Gray = Light segment, the bottom half with low monthly online usage;
•    Primary groups = the heaviest online users in terms of usage intensity;
•    Secondary groups = exhibits some of the same characteristics as the primary segments, but less online intensive (fewer minutes + page views per month).

While there’s clearly a lot of data to digest, this chart doesn’t show the relative size of each or usage intensity of each segment. To put that into context, we created a visual representation of the segmentation matrix to show each segment’s size as a % of the total (#2, top) and online usage intensity in % of total minutes (#3, bottom).

US Internet segments, source = comScore July 2008

US Internet Usage % Minutes, source = comScore July 2008

Some Highlights and Interesting Takeaways:
•    The top 20% of heavy US Internet users make up 69% of online time spent.
•    75% of heavy users are NOT using Firefox [Primary Acquisition segment].
•    Primary acquisition segment is approx. as large as the Firefox base (29 Million unique users), but has a different profile. (More on that later). Also, note this heavy user segment is 15% of the population that represents over half of the usage– wow!
•    Primary engagement segment (4.1M unique users) are dual browser users who use Firefox but spent the majority of their online time using IE.
•    Of Firefox users, nearly one-third are heavy Internet users and represent 78% of the total time spent on Firefox.
•    The primary retention segment (approx 4.9 Million unique users) or 3% of the total US Internet population drove 10% of the total online minutes.
•    Current Firefox users range from light to heavy Internet users, while about half are light Internet users.

Remember, the objective of this segmentation project is to help us think about how to action insights to better inform our marketing. Rather than boil the ocean and develop a set of strategies for each group, we decided to focus on the three primary segments (right side of the chart in the red box) that collectively represent the top 20% of US Internet users.

In my next post, I’ll get into more detail about each of these primary segments. Meanwhile, I would welcome your thoughts and ideas about things the marketing team should consider to either help acquire new users, engage those secondary Firefox users, or retain our most loyal users.