Skip to Content

Making sense of complex data

Within research, it’s easy to get overwhelmed and intimidated by new terminology, which can often feel like jargon. There are several buzzwords being thrown around these days: Data Science, Big Data, Machine Learning, to name a few. But what do they all really mean?

Caroline Dieleman Senior Director 26 October 2022
Segmentation might be tricky but making sense of data doesn’t have to be.

You may have already heard things like: ‘anything that does not fit in an Excel file is Big Data’ or that ‘Machine Learning means that you don’t need to write equations out by hand’. As it stands, however, there are no official parameters for what is and isn’t included within these terms.

Whether you use “Big Data” or “Machine Learning” can differ massively depending on the objective set out. What helps is to start from this objective, break it down into pieces, and work your way backwards.

For example: A company wants to create a segmentation to target their customers better.

Step 1

CRM data audit: What information do you hold in your CRM?

  • How much confidence do you have in this information?
  • When was it last updated?
  • What is the data structure?

Step 2

Research: Create a market study with a boost of current customers. Mirror the relevant questions from the CRM in the research as well as additional “why” questions such as values, beliefs and attitudes.

Step 3

Append 3rd party data to both your CRM and research (ACORN, Experian, to name a few).

Step 4

Create the segments with a mix of behavioural questions from the CRM and attitudinal questions from the research. A fine balance needs to be kept: if you move too far to the behavioural side and the segments lose their usability (the “why”). If you move too far to the attitudinal side, you will lose their findability (their link to the CRM).

Step 5

Create an algorithm to identify the CRM questions that best predict the segments current customers are in.

Step 6

Apply the algorithm to the whole CRM, creating segments for your entire customer base.

These might seem like 6 easy steps, but there are hurdles at each step of the way.

Some of the barriers we have seen are:

  • Internal politics: If you have separate data and insight teams, they will need to work together to work on the data audit, understanding that insight will enhance their data.
  • Stakeholder engagement: segmentation is an iterative process, key stakeholders in the company need to not just understand the building blocks of the segmentation but also be bought into the solution to be able to sell it internally.
  • Single focus: research in general has the tendency to want to deliver the moon on a stick, especially when there are multiple stakeholders involved. This approach is specifically dangerous when creating a segmentation as you need a single focus, or you will have no focus at all and will deliver on none of the objectives.
  • Time/Investment: There are no shortcuts in this type of work. This means that it will take time to go through each of the steps properly.

There is a lot of noise out there, so it can be hard to know what is fact, and what is fad. It is easy to get lost in the world of shiny tools and terms, but as long as you bring it back to your research objective and keep asking “How can this help me get towards my goals?”, you will be able to cut through the noise.

Segmentation might be tricky but making sense of data doesn’t have to be – read more on advanced data analytics here or start a conversation with our experts here.