Month: July 2019

Blog Customer Segmentation Use Cases

Customer Segmentation & Personalisation with Machine Learning


Customer segmentation is a method of dividing customers into groups based on their shared characteristics such as age, gender, interests, or purchase habits. This is important for allocating company resources and targeting clients effectively. Customer segmentation influences mainly sales and marketing decisions in organisations. It allows marketers to tailor their campaigns to different target groups for promotional, marketing, sales, and product development strategies. Personalised marketing and sales efforts that are based on well segmented customer groups have much higher conversion rates and therefore impact an increase in revenue. 



Even though customer segmentation is such a vital part of organisational decision-making and machine learning is advancing rapidly, very few companies are using automation processes to optimise their efforts in customer segmentation. Organisations that are using machine learning algorithms in their customer segmentation efforts often use “clustering”. 

Clustering is a machine learning technique for identifying and grouping similar data points (e.g. customers) together. The objective of clustering algorithms is to ensure that the data-points in the same group are very similar, and data-points in different groups are extremely dissimilar. This process is “the hard part” that  42% of marketers in the USA and UK do not complete due to lack of employees, time, and funding, which prevents them from reaching their personalisation goals.  


Once customers, stores, or locations have been segmented into groups, managers can make decisions about how to target each group most appropriately based on their differences. For example, specific segments may be optimal targets for inbound marketing campaigns such as cross-selling and others may be better suited for outbound campaigns such as email messaging. Thanks to machine learning models, companies no longer need to spend ages defining segments and allocating customers to them. Now, their time can be used for understanding who will be best served through which channel, with what message, and at what point in time. Companies can now focus on how to maximise the impact of current capabilities and resources such as available channels, agents, content, and offers!  


In addition to simply automating the grouping process, machine learning can impact the “contacting” process. Machine learning models will learn (based on each customer interaction) whether the chosen communication channel, content, offer, and the point in time were “right” for a given customer. Based on the customer interaction with campaign touch points, machine learning models can automatically predict what engagement mode is most likely to delight each individual customer. The more data there is, the more accurate the models will be. Machine learning models do not even need to base their assessments on data from current campaigns – historical data from customer engagement with previous campaigns can be used to train the models.  



Other Benefits of Doing Customer Segmentation with Machine Learning 


Simpler & More Precise Segments  

These days customers cannot only be segmented based on their demographic or purchase behaviour data for effective segmentation results. These segments are simply too broad and deliver only suboptimal conversion rates. Organisations are now including more advanced additional criteria to ensure better segmented groups. Specific data from the customers’ sales journey is included such as whether they attended an event, clicked through to the company website, clicked on a product, and whether they bought the product or not. The desired criteria to include is always changing and often growing. If managed only by humans, it can quickly become unmanageable and inaccurate. Allowing machine learning models to do this process for you, ensures your customer segments are always optimised, saving employee time and reducing the organisation’s costs.   


Changing Segments 

Decisions to change the segmentation criteria to include different characteristics have perviously been a challenge. Manually reallocating customers from their initial segment to another requires constant monitoring and shifting. Machine learning models automate this process and update and shift customers from one group to a more relevant group in real-time. In essence, the customer segmentation process does not need to be time-consuming, expensive, and manually adjusted, thanks to the implementation of machine learning clustering in this process.


Concluding Thoughts 

There is no longer a need to manually allocate customers to segments. With machine learning, each customer is their own segment, defined by any chosen criteria that organisations deem best. When it comes to customer segmentation, machine learning algorithms are highly effective for personalising the customer experience and leading to conversions. 


If you want to learn more about how Machine Learning can best be applied in your organisation, contact us to find out you could be saving time, money, and resources.  

Blog Use Cases

Supply Chain Optimisation & Machine Learning

Although it is not a media-attention-grabbing industry, the supply chain and logistics management industries have been in leading machine learning development industries over the last ten years. New technical developments have allowed supply chain and logistics organisations to leverage their data in ways that is decreasing expensive machinery failures, exceeding customer product and service expectations, and generally increasing their long-term return on investment.  


It is critical to have solid foundations when it comes to supply chain planning – one of the most important activities in supply chain management. The implementation of machine learning in the supply chain decision-making process has allowed for significant optimisation. Supply chain teams are now able to balance demand and supply and optimise the delivery process due to machine learning data analysis and intelligent algorithms. Machine learning in supply chain and logistics management decreases the amount of manual work that needs to be done by humans – large data sets can be automatically analysed without any natural human error. 


Here are a few of the areas of supply chain and logistics management where machine learning models are being implemented for optimisation: 

Inventory Management 

Both under-stocking and over-stocking are common problems in supply chain management that can ruin even the most efficient supply chain strategy. Machine learning algorithms can monitor the stock levels in a warehouse and take into account more factors than typical forecasts – this can improve inventory optimisation, particularly in businesses with multiple distribution locations. This is useful for reducing transport costs (transporting a more accurate amount of stock, fewer times) and keeping inventory at a “comfortable” level (idle stock levels can be decreased based on more accurate machine learning predictions of future demand). By analysing current inventory data, inventory management can be improved to ensure optimal business performance and allow workers more time to focus on product quality and customer experience. 

 Warehouse Analysis 

Machine learning has enabled computers to “see” and understand images and videos – this is called Computer Vision (CV). This tool enables warehouse automation. Product barcodes can be processed, warehouse perimeters can be monitored, and employee activities can be followed: preventing safety violations and theft or trespassing.  


In addition to CV, machine learning is enabling robots to take on mundane tasks that would usually be assigned to a human. For example, machine learning robots can pick ready goods to send to customers, can carry up to 500kgs as they make their way around with sensors to avoid collisions, and can be summoned by workers at any time. This technology is improving team productivity and efficiency in the warehouse while reducing the risk of human injury (from lifting, managing, and carrying heavy goods) and human error – like a parcel being sent to the wrong address.  

Logistics Route Optimisation 

Customers expect their deliveries to arrive at the time and date promised to them by the company. Therefore, it is important to ensure this happens without any delays and in the most cost-efficient way possible. Delivering parcels to customers on time is no simple task – for example, delivering just 25 parcels by van has 15 septillion (a trillion trillion) route options. Machine learning models can analyse existing route options and implement route optimisation – creating a current optimal route depending on road conditions, weather, and other factors. This means companies are able to reach bigger geographic locations, reach more customers efficiently, and generate more revenue. Of course, keeping the customer satisfied too. 

Supplier Relationship Management 

When you make an incorrect decision about your organisation, your business can suffer. This also applies when making decisions about your suppliers (in the worst case, a bad supplier choice can end your business). Machine learning can make the task of selecting a supplier and maintaining a relationship a lot easier. Reliable predictions about almost every interaction with an existing or potential supplier can be provided by machine learning models based on historical data sets (such as messaging records, audits, or credit scores). In this case, machine learning can help organisations avoid selecting an untrustworthy supplier and maintain a long term mutually beneficial relationship. 

Workforce Planning 

Most industries and business sectors involve team management and planning. The same goes for the supply chain and logistics management industries. Team management processes involve recruitment, retention, employee training, sub-team management, and performance analysis and more. Machine learning solutions can be applied to these areas of workforce management. For example, machine learning models can identify key traits of the most successful employees and use that information to shortlist most ideal candidates form a collection of CVs. 


Research shows that employees leave their jobs due to salary less often than expected. This means that employee turnover can be decreased without increasing salaries. Machine learning predictive analytics can save companies money on hiring expenses, training costs, and overall payroll expense by leveraging employee data (employee surveys, for example). This data could provide insight as to what really makes employees happy at work. This can influence a positive work culture and increase productivity without breaking the bank. 

 Your Article Has Been Delivered! 

In conclusion, machine learning has applications across multiple sectors in supply chain and logistics management. From ensuring inventory levels are optimised throughout changing demand levels over the year, to monitoring warehouse and employee safety, to moving inventory in the warehouse more efficiently, to picking the most optimal route for product delivery, selecting and maintaining a healthy supplier relationship, and to team management, machine learning can leverage your organisational data and allow employees more time to focus on customer satisfaction. Allow machine learning models to increase your business revenue and be a long-term investment that save time, resources, and costs in your supply chain or logistics management organisation now.  


Maintain your competitive advantage and contact Intellify for more information on how machine learning can be customised for your company to allow you to optimise all department involved in your product or service delivering.  

Blog Events

PyData Sydney – July 2019



This month, PyData was hosted by Intellify at the very slick Executive Centre in Sydney’s CBD. Although seats were limited, the room filled up quickly and was bustling with chatter in only fifteen minutes.


Soft drinks and pizza were enjoyed by all and the crowd of data scientists, analysts, and other machine learning enthusiasts did not need any encouragement to get their networking going!

The first presentation of the evening was from Senior Data Scientist Ammar Mohemmed. He gave an introduction into Optimisation using Python. His presentation was technical as well as practical – with lots of useful examples from real-life scenarios where price and quantity were involved. His presentation made clear that while machine learning and artificial intelligence technology is an absolute time saver and more accurate than manual human calculations, there will always be a need for human analysis from a business perspective to re-check the suggestions that the models generate.



The second presentation was from another data scientist from Intellify, Jordan Wade. He presented an algorithm that combines both genetics and neural networks – called NEAT (NeuroEvolution of Augmenting Topology). He enlightened us on how neural networks in artificial intelligence can evolve in a similar way that biological life does (except the technology is much more sophisticated in many ways). Evolutionary algorithms always heavily mirror biology, neuroevolution being no different in this respect.


The presenters answered questions and conversed with members of the PyData community and the evening concluded.


We would like to thank our venue host and all the members of the Pydata community who joined July’s meetup. We hope to see more of you next time!

For more information about Pydata and our events please click here. Follow us for more events where you can hear from practicing data scientists on their experiences machine learning methods. Proudly hosted by Pydata Sydney’s sponsor – Intellify the 2019 AWS Partner of the Year in Data, Machine Learning and Analytics.