Segmentation Analysis Making Sense of Initial Event to Ultimate Outcome
I love segmentation analysis. It gives a deeper understanding about the consumer behaviors and the true value of marketing efforts. It gives a better sense of what’s happening to the outcome rather than the vanity traffic metrics that many marketers or managements love.
People running online business knows many marketing efforts don’t immediately tie to sales. Marketers fuzzily know that one day those consumers who were exposed to the ad have the brand awareness and will one day buy from your site, or else where, or even share the product info to their friends. Let’s try to cut away from that fuzziness.
Let me quickly talk about social gaming… Many social gaming companies track DAU (daily active users) or MAU (monthly active users) as KPI, because overall number of users really don’t matter when they need to make money and growth through active gamers playing their game on their platform. In most cases, those games are usually free, and as users get addicted to playing it, gaming companies would do their best to drive converting those gamers to paying gamers. They pay either through monthly subscriptions, avatars, unlocking stuff, get more stages, remove ads, etc.
In such case, cohort analysis comes into picture in their reporting and analysis practices. A cohort study or panel study is a form of longitudinal study used commonly in medicine, social science, ecology, etc. In this case, it is an analysis of factors and follows a group of people who were exposed to a particular event and tracked across time to see how their behavior change. A cohort is a group of people who share a common characteristic or experience within a defined period (e.g. downloaded the game for the first time, signed up for an email, create their account). Then this cohorts will be followed across time to see if they perform the next desirable action.
Here is an example snap shot from KissMetrics.
In this particular example, it is showing the number of people signed up during that reporting month. Then it follows that group of people in month 1, 2, 3, and so forth measuring “Signed In”. So in month March 2011, there were 1,312 people who signed up for this service, and 30% of them have came back and signed in 1 month after their initial sign up. Seems like that 30% was the all time high when compared to historical. That could indicate some correlations to some marketing activities to drive that conversion.
To me this is a great example of how powerful metrics could be to articulate the effectiveness of marketing or any business efforts to drive the outcome. Here are two ideas I randomly came up with you could do:
- Track the cohorts of Marketing Campaign XYZ on people who had been exposed to the ads for the first time. Follow them and see when they convert, or sign up for newsletter, or share content. See how changing marketing tactics or optimized ads impact the recency of the outcome.
- Track the cohorts of people who have converted to sales on your site for the first time. Then follow that cohorts to see if they come back to re-purchase something else of a higher priced items.
This analysis is pretty hard to do with traditional web anlaytics along as cohorts are defined as particular group of people who you explicitly know. In other words, if people delete their cookies or you loose track of the original cohorts then data becomes pretty dirty after several months. However, if your registrations, logins, sign up events are tied to consumer info, then you should have less challenges in tracking them.
Perhaps you can trying this with your CRM database first, and see how that come out to.
Next… what about some deep dive segmentation in web analytics…
This is not a cohort analysis, but Google Analytics advance segments could give you a pretty good deep analysis of your converting audience. It looks nothing like the cohorts analysis table, but that’s fine. You’re looking for insights and not for pretty graphs/charts.
Taking a look at my own traffic for some random period. I want to compare the new visitors traffic who came from my feeds via feedburner, and see how that compares to returning visitors from the feeds. You could compare the outcomes, and two different segments provide different results, which is great, because I know now that my returning audience through feeds are super precious. Their goal conversion rates, and AdSense response rate and eCPM are much better than my new visitors through feeds. That means I must be doing well bringing back engaged audience and they’re converting.
Now you can look at this return visitors segment’s recency report and see how recent they visit, analyze and get a good sense to see if that recency is shrinking, is it aligned to blog posts recency, etc.
To me this is the important thing that comes out of this is the “story”. There are many marketing tactics that drives spikes, and there are many common metrics that measure that impact for that moment. A lot of times we don’t know how to replicate that… Not so good… Many analysts and marketers focus too much on that. These insights or anlaysis tactics I showed you should tell you what drives the growth of your key audience, or what marketing efforts could be replicated to bring more of these lucrative audience.
- Planning for Tag Management System TMS Implementation
- How to use heatmap studies to build good hypothesis and increase task completions
- Important planning process and methodology for A/B testing
- What is distracting your customers from your website
- Digital Analytics Roadmap 2013
- Page Load Speed Impact to Bounce Rate and Value
- Using visitor goal, department goal, and analytics in optimization planning
- Screening through mobile metrics
- Generating awareness from YouTube and digital strategy expectations
- Key considerations for a landing page optimization