Search

15 . 9 . 16

Data Driven Decisions


Data visualisation, modelling & forecasting are all hot topics at the moment, but what does all this mean and what insight can be driven from the data you are collecting?

The first hurdle is identifying the right data to collect, whilst ensuring this is enough to answer the questions you are trying to tackle – without ending up with a stockpile of meaningless data. Yes, sometimes less is more!

There is the planning aspect in which all cross departmental business objectives and KPIs must be obtained. This is a good place to start with collecting the correct data, ensuring all parties can have the data they need in order to measure what is important to them. This is probably the most important stage, as this defines the data output and therefore the analysis and conclusions drawn.

Following this there is the implementation stage, regression and validation steps, and finally the reporting and analysis stages. We will mainly be focusing on the final stages in this post.

Types of Data

Often we are looking at quantitative data from a web analytics product such as Google Analytics. However we are increasingly integrating some of this information into CRM systems to expand the level of insight and understanding for entire customer journeys. In addition to the above we are more frequently reviewing data from surveys and other user feedback methods, often qualitative data, to further analyse to prove or disprove hypotheses.

webid

Basic data analysis can, in the example of qualitative data, show the number of users who responded to a question and perhaps a grouped breakdown of their responses. However, what about the segment of users who are in a certain age group who purchased your product? Or those that had used one of your competitors previously? Are these users significantly different from another segment? For web analytics data we may want to look at specifically the users who became a customer, what did they do prior to this? Is it where they land? Where they go? Etc. etc. Before we know it we have so many open questions to consider. With so much data collected, what is important and where do we start with analysis and insight?

datacapture1

Our Approach

Here within the Twentysix data team, one of the tools we use is R. R is an open source statistics program to better manage and visualise multiple datasets. R also has many open source packages that can be downloaded and integrated with the base system. Common packages we currently utilise are xlsx and RGoogleAnalytics.

These packages allow read and write functionality from Excel to R, enabling us to work better with client datasets and also produce outputs suitable for a wider audience. RGoogleAnalytics integrates R with the Google Analytics core reporting API, allowing us to query this data and analyse & visualise this directly through R.

Many advantages to using the Google Analytics interface include:

  • Scripted data extraction that can be saved and re-run
  • Ability to pull more than 10,000 rows of data in batches
  • Mitigate query sampling by rapidly splitting date range queries
  • More scalable customised reporting and the ability to view more than two dimensions against many metrics and complex segmentation.

We also have a purpose built powerful computer we can connect to remotely to run queries and analyse some of these complex data sets. For example our attribution modelling reviews each touch point for each purchase journey and applies a weighted value to each, dependent upon frequency, position and path length, amongst many other factors. Our latest survey analysed contained almost 20,000 respondents to over 30 different questions spanning both qualitative and quantitative data.

In order for these types of data to be processed, visualised and for conclusions to be drawn, the easier management, visualisation and quicker analysis and processing capabilities of R significantly reduce the time spent on data collection and manipulation. The machine can focus upon this whilst we have increased time now available for analysis and understanding of what this is telling us, and what to do next.

Some examples of our recent R graphical outputs are shown below:

Figure A: Scatter plot showing interest & knowledge of product for different age groups. Figure A: Scatter plot showing interest & knowledge of product for different age groups.

 

Figure B: Scatter plot showing line of best fit trend and expected variance over time for registration completion data Figure B: Scatter plot showing line of best fit trend and expected variance over time for registration completion data.

 

Using data to see what is or what was happening to make conclusions and drive strategy is an extremely powerful, impactful and intelligent solution that we can use to help grow your business.

Want to find out more about what Analytics & CRO can do for your business? Contact us here.

Phil Gawrylo is Senior Data & Insights Analyst in the Conversion Optimisation team at twentysix. He enjoys the challenge of unearthing actionable insights from data analysis to help inform the strategy of our conversion rate optimisation campaigns.

Share This:


Comments