Uncover your best traffic sources in 45 seconds
Google Analytics and other “classic” web analytics are great at tracking data but a bit less efficient at giving actionable insights.
Let’s take a look at the Acquisition report from Google.
What do we have?
- Highest providers of visitors (number of sessions per sources)
- Some behavioral data (bounce rate, pages/session, etc.)
What kind of measures can I take? Nothing. It’s only emphasizing my current state. Of course I can correlate this with other reports. As a web marketer expert that’s my day to day job. But let’s face it, we are wasting time.
Now let’s take a look at the same data through a behavioral analysis.
I’m using my new toy: Behaviors Finder by mQment. It’s currently in early access at mqment.com/behaviors-finder-beta.
To begin, I connected my Google Analytics as my datasource. Then I ran the Acquisition report from the dashboard page. The app collects data from the last 30 days and runs a clustering algorithm to “group” similar behaviors together. Results are shown below:
For my example, I defined a successful traffic source as a mix between ratio of new sessions and page per sessions.
In other words I want to uncover traffic sources bringing new visitors that are truly interested in my offer.
To do so, I look at the engagement score. It’s a neat feature related to the chart axes. In my case, it’s calculated from Axis Y with “% New Sessions” and axis X with“page/sessions”.
I also added the bounce rate metric as the radius. A smaller circle means a better engagement with our website. Now the chart becomes very visual. On the upper right is what I call the success area. Groups here have an ideal behavior.
We see that B has a successful behavior while D is terrible. We will compare them and see the differences.
Segment B is a strong winner with a lot of new sessions and a low bounce rate. Something we’d like to see for most of our traffic sources. On average there is also way more pages seen per session.
All theses metrics are interesting but nothing actionable yet. It would already be actionable if the behavioral analysis was based on event typed data (Mixpanel, Heap Analytics, Amplitude, etc.). More on this in another post.
If we scroll down a bit, we see the traffic sources with a behavior similar to the selected segment. That’s where our actionable insights are for this report.
I blurred all the sensitive data. Next to each entries you have the Similarity score. It tells, on a scale from 0 to 100, how visitors from this source behave like the selected segment. 100 is best.
For this client I was able to identify well perfoming and underperforming referrals at the same time: we had bloggers driving a lot of low quality visits. We used to pay them, not anymore. We found new partners and redirected our budget to the best referrals. Today we have less traffic but our revenue has grown by 15%.
We are opening access to our beta, feel free to try at mqment.com/behaviors-finder-beta. You will be able to connect your Google Analytics and Mixpanel account. Share your feelings and discoveries in the comments!