5 Pieces of Advice for Value-Adding Data Science

The value-adding data science is the work you see on your bottom line; it improves conversion rates, retention rates and/or your revenue per customer. As in so many aspects of life, value-added data science demands prioritisation of time, money, and – if you do not have these competencies in-house – a good business partner. We provide you 5 pieces of advice to value-added data science based on the 5 typical pit falls identified in our recent study of data collection and use among B2C companies

 

Share article:

1. Your Data Is Not Perceived As Valid

Your customers data’s value in your segmentation project is proportionally increasing with their validity. Therefore, you should make sure that master data like your customers’ names, addresses, and phone numbers are updated and validated. To put this in perspective using a DK reference, last year, more than 900.000 people changed their addresses in Denmark alone, and more than 600.000 Danes got a new phone number. If your customer data have not been updated recently, it is a serious pitfall for your segmentation project: potentially, not even your geographical customer spread is fair

Fix 1

Update and validate your data immediately before your segmentation project

Having your data “washed” is updating and validating your customer master data and gives you a good foundation for your segmentation project; it strengthens your analytical validity (e.g. geographical spread) and ensures your customer data can be blended with other data to give your analysis more depth

Learn more about our solutions for master data management

Fix 2

Validata data from the beginning

 

An ambitious and value-maximising segmentation project is an ongoing process where algorithms and segmentation criteria and continuously evolving with your business. Therefore, validating your customer data from the beginning is a good investment

Learn more about how we can help you with fast and easy access to validated data

Fix 3

Establish a data management procedure

Even if data is validated when collected, they will become obsolete over time. Therefore, at continuous update of your data is the way to ensure your also continuous segmentation model updates are founded on the most valid data. Besides, it is also a part of your GDPR compliance to keep your customer data updated and relevant

2. You Use Too Many or Too Few Data

The art of data science is choosing the right data: the data blend that gives you the insight that will drive your business forward. That is why it is important that the data you use is relevant and can create the necessary customer insight

 

If your customer data consists of a rich mix of master data with names, addresses, gender, age, transaction data, channel interactions, etc,; then you have a good foundation to get to know your customers insights & needs, their RFM score or value, and you can then match these value groups to the products they buy

 

But you cannot explain why

Fix 1

Use more external data

That knowledge you get by including more external data. Even though more data in itself will now lead to the wholly grail, the inclusion of the right extra variable will provide you with a full 360° understanding of your customers. Maybe their purchases are affected by their housing type, income level, family status, behavioural needs or other dimensions yet undisclosed in your data science, if you are only working with your internal data

In our data catalogues you find an overview of the variables most often requested by our customers and ideal for supplementing your internal customer data

Fix 2

Identify the explanatory variables

To identify the explanatory variables can be difficult and when done manually by analysts, important points might be missed. That is why everyone is talking about machine learning; with the right algorithm, previously unseen patterns are disclosed, and your segmentation adjusted accordingly. Based on our extensive experience within working with B2C companies throughout the Nordic region, we have developed our own machine learning algorithm engines that both identify the variables most explanatory in your customers’ need – e.g. churn, and at the same time help predict the next event

Learn more about our approach to segmentation

Fix 3

Eliminate redundant variables

If you use too many variables, your data pool becomes unmanageable. The right amount of data is largely dependent on the purpose of your segmentation, so tailor it to your analytical question. If you identified the explanatory variables as described above, the redundant variables are easier to identify. And if you do have redundant data a positive side effect of cutting them loose is that you become more compliant GDPR concept of data minimisation

3. You Lack Data Science Competencies

Setting up the right model is not simple. Identifying the right questions that the segmentation should answer is a task that requires a good overview. Therefore, if your company is inexperienced within customer analysis and segmentation, the segmentation can potentially cost immense resources without creating the expected value

Fix 1

Engage data scientists

Data science is a discipline mastered by professionals with an in-depth knowledge on how data should be structured, analysed and interpreted to support your business strategic goals. Engage experts to get yourself on the right road

Fix 2

Use an analytical tool that matches your need

Your analytical competencies are also restrained by the methods and software you use. The range goes from Excel to actual analytical tools that can handle machine learning, AI, and visualisation. There are many solutions out there so invest in a tool that matches your needs. Your research could start here: 2018 Gartner Magic Quadrant for Business Intelligence and Analytics

Let Us Help You Create Value-Adding Data Science

Optimal data usage enables you to realise your business’s potential. We help you to identify the right data to make your organisation more effective and optimal. We systemise the usage of smart data within your customer database, identifying new connections, new understandings, and new target audiences


Data based decisions are important, when your objective is efficiency and precision. We have worked with register data for more than 16 years, and we have helped many organisations choose the right data sources and variables to create useful analyses

 

keyboard_arrow_rightLearn more about Geomatic Insights

4. Your Segmentation Model Is Irrelevant

Almost a third of the survey respondents in our recent study of B2C companies' data collection and use replied that they do not know where their segmentation model originates from! It is a basic condition that the rooting of your segmentation model is perceived as relevant by all employees. If the do not, each department will develop their own or use nothing at all, thereby ensuring your investment will not result in improved sales and retention

Fix 1

Document your segmentation model

Whether your segmentation model is developed in-house or rooted in a standardised segmentation tool, its origin should be documented and described in a way that ensures that maintenance and further development is made independent of changes in staff

Fix 2

Make sure your segmentation is based on data

Your segmentation should be based on data from either your existing customer base or your research of a potential market. Personas based on brainstorms and wish lists are almost always impossible to operationalise

5. You Lack Ambition

If you – like 46% of the survey respondents in our study – think your data science is high-level, but simultaneously do not find your segmentation model is adding value it is time to raise ambitions. To raise the level of ambitions of one’s data science is not equivalent to change the segmentation model if it is already good. But maybe it should be implemented better – or its validity would increase if the data were better or updated? Either way it is not enough to wish for value-adding segmentation; it takes time, money, and – if you do not have these competencies in-house – a good business partner

Data Science that Drive Your Business Forward

Sign up or get in touch with Oliver Newton directly to learn more about our data science and approach to segmentation

Oliver Newton portrait

Oliver Newton

Key Account Manager

+45 5183 9125

Linkedin