You've probably seen us talking a great deal about how our data analytics services can help you, but you may not know that we use our own services as well! On a regular basis, our analytics team brainstorms and experiments on Construct’s data, and if it works well, we bring it to our clients.
Today, we will cover how we turned parts of Google Analytics data into actionable market intelligence for our internal marketing team.
What is Google Analytics?
Google Analytics, known in the industry as GA, is a free “service that provides statistics and basic analytical tools for search engine optimisation and marketing purposes” (source). There is a wealth of information within GA; with tabs that lead into menus that lead into sub-menus, this can feel overwhelming.
Since we deal mostly with customer analytics, we get most of our data from the “Audience” and “Behaviour” tabs. Using this data, we can help our clients identify and segment their customers, figure out where their leads come from, and build propensity models based on behaviour (among many other things).
For this internal campaign, we wanted to explore mobile device data in particular, so we analysed data from the Audience -> Mobile menu.
Mobile Device Intelligence
Before we dove into the data, we had a few questions we wanted to answer:
- Which mobile device companies hold the largest market-share in Singapore?
- How much are Singaporeans investing on their mobile devices?
- What screen resolution is most prevalent? What screen size?
So, what? You can train anyone that knows their way around a keyboard and the internet how to get this information; the real value comes from the time we save through automating the process and knowing what information and intelligence you can derive from the information.
Using this information, here are some examples of the intelligence we can gather:
- As an app developer with the bandwidth to only design for one app market, should you design for the App Store or Google Play Store?
- For a long time, 1024 x 768 was the gold standard for basic website design (source 1, source 2). Is this still the case? Does this hold for Singapore?
- The smartphone industry continues to produce larger phones each year. Is there truly a demand for this? Are people opting for these large phones?
Our Data Methodology
In order to ensure we could generalise any conclusions we made to the population, we needed as many data points as possible. To accomplish this, we pulled numbers across a 6-month timeframe from our data, giving us millions of sample points and hundreds of mobile devices.
In addition to the GA data, we needed to supplement this data with details about phone screen size, screen resolution, and retail prices. We got this information from the mobile devices from industry leader GSMArena. Recall from our last blog post on ethical web scraping that we should check if we are allowed to scrape this pricing information. After verifying from their robots.txt page that we could mine their site data, we ran a Python script to obtain the prices we needed.
Once we had enriched this information, we needed to assign weights to balance out differences in quantity between sources and devices. After detailed research, we utilised a technique called stratified sampling to achieve this. For a thorough explanation of how we did this as well as a copy of our Python script, please refer to our GitHub page.
With the weighted numbers, we visualised everything on a downloadable infographic content and an online Tableau, which you can view in our downloadable content here.
Data @ Construct
This is just a small glimpse into the types of analysis and insights we can derive from Google Analytics. As we mentioned earlier, we can combine this data with other data from GA or other sources and derive further insights or build predictive models that can help improve campaign performance and increase revenue.
To find out more about other ways GA can help your business, or how we can help you solve some of your other problems (such as creating and managing your data pipeline, deriving insights from customer data, and predicting sales decisions from your sales data), drop us a line for a chat. We’re friendly. Promise.