Back to all articles

Use customer segmentation to create better customer experiences

Rick van Sintmaartensdijk
Rick van Sintmaartensdijk
Data Scientist
6 min read
1 April 2021

Most industries from retail to B2B want to segment their customers to better understand their needs and behaviour. However, nowadays, a typical customer journey spans multiple platforms and can include both online and offline interactions. To stand out from the crowd, brands need to be able to harness the data at their disposal and segment consumer information to create unique and personalised experiences for their customers. For many companies, this feels easier said than done. But it doesn’t have to be this way, below are three steps to help you get started.

Start by centralising your data

So how can you start segmenting your data? Begin by gathering data from both online and offline sources and pool this information together into a data warehouse, also known as a Customer Data Platform (CDP). Now, numerous cloud platforms such as Google Cloud Platform, Azure or AWS, can be used for this purpose. With data engineering practices in place, data can be pulled into these platforms from various sources such as your CRM, media platform data or POS system. The information is then stitched together to create a single customer profile. This unified customer database creates a holistic image of individual customers and visitors. The data platform is flexible, scalable and makes the information accessible to other marketing systems.

Data-driven customer segmentation

However, simply having all your data stored in one central spot is not enough, but it does open up endless possibilities to enrich and activate it. For example, you can use the available data for customer segmentation. However, when you are implementing this, you might face the following challenges:

  • Which specific data points should one use?
  • How many segments should one aim for?
  • How could I use these segments to improve business results and customer experiences?

And these challenges are extremely difficult to tackle if you are using market reports with general personas or classic rule-based segmentation. But, these become manageable when one opens up to data-driven techniques which can help brands overcome these challenges. These techniques use your data to generate insights that are difficult to derive when using rule-based segmentation or general personas as a starting point.

One of these techniques is called clustering, which will help companies find the optimal number of naturally existing groups in their data. The data points fed to the clustering algorithm should be selected based on a combination of data expertise and custom business expertise. The combination of these two inputs often solves the challenge of selecting the right data points, it ensures that you select data points that are relevant for the business and for the algorithms. The result is a persona assigned to every individual customer, which makes it usable for activations on your platform or in media.

Using the result and the characteristics across the data points per segment, interpretation and the creation of data-driven ‘personas’ becomes possible. These personas are your gateway to knowledge on your individual customers type of behaviour and needs.

Now, the technique of clustering needs to be applied only once, however it will not continue to assign personas to any new customer data. Therefore, brands can harness the power of predictive machine learning to assign a persona to every new customer. The model will learn the ‘rules” which have been put in place by the clustering algorithm and is able to apply them to every new input in real-time. By having the most up-to-date information about your customers at your fingertips, brands can ensure their marketing budgets are being properly allocated.

Using segmentation to activate your data

Once your data has been centralised and enriched, it can be used for activation purposes. This means implementing the changes you learned by segmenting your data into your own and paid media platforms. Leading you to address customers and visitors based on their behavioural characteristics. 

To find out how to best communicate with different audience persona’s, brands can A/B test which message resonates the most. For example, what message could you show to a customer that often buy products for the entire family rather than simply themselves? Alternatively, how could you best communicate with a customer that does a lot of online research before coming to your store to buy a product? To determine this, brands can tweak their messaging on their websites or in emails. By having these personas as your disposal, brands can hyper-personalise their content according to customer behaviour to provide customers with the best experience. 

Brands can also adjust CTA’s and bid strategies based on persona characteristics. By creating look-a-like audiences of your most valuable personas, you can target and attract customers who exhibit similar characteristics.

The key to success is to experiment. The data tells you what each individual customer is typically like, the only thing left to do is create a message which suits each individual customer. If you achieve this, your brand can increase its website visitors and revenue as you will be reaching the most relevant target audience.

Use your data to stand out from the crowd

By segmenting one’s data, brands can better understand their customers wants and needs. Though the process may appear complex, it’s fairly simple: gather all of your data into one digital container, then use data-driven techniques and machine learning algorithms to enrich that data and activate it so that your marketing systems can use it. Also, don’t hesitate to start small and apply this model only to web behavioural analytics, for example, or CRM data. If you see positive results then you can increase the amount of data you include in this model. By harnessing your data, your brand can better cater to your customer’s needs and can personalise your media buys for optimum and relevant reach.

More Insights?

View all Insights


Data Scientist

Rick van Sintmaartensdijk