Unsupervised Machine Learning

Unsupervised machine learning is used when you want your system to identify patterns in your data and infer trends. It can be used to find patterns and trends that humans might not be able to see. Unsupervised machine learning is a self-learning system that will figure out patterns in your data that are important.

Insight from a Head of Growth

Why choose unsupervised machine learning?
Saas customer churn rate is consistently over 20% and no one knows why.

Meet Adam
Adam is the Head of Growth at a B2B SaaS company that was looking to reduce its customer churn rate.

The blind leak
For two years Adam’s organisation has been suffering from an excessive customer churn rate. The difficulty is multidimensional. They don’t know exactly why so many customers are leaving. They cannot predict which of their current customers are most in danger of leaving. And even if they did knew who to target, there’s no clear intervention that would change their mind.

This is a classic forest problem. To understand what’s happening, imagine that the customers about to cancel their accounts are lemmings heading towards a cliff. Meanwhile, the satisfied customers are placid deer moving through the woods.

Adam’s team operates up in the clouds. They looked down at a thick forest of data trying to identify the lemmings that are about to hop off the cliff.

Headless lemmings
The biggest problem is a lack of information making it difficult for the marketing and customer success teams to collaborate. 3,000 metres below is the customer success team. This team runs helter-skelter through the woods trying to catch lemmings. They react to advice from the marketing team on where to find lemmings. But, often when they arrive expecting frustrated lemmings they find happy deer. Meanwhile, elsewhere in the forest, more lemmings meet their fluffy end.

Increasing visibility in the woods
Adam’s team needed a layer between marketing and customer success. This layer identifies lemmings early enough to give the customer success team time to arrive and coax them from the edge. They might even be able to convert them into content Bambis that would never dream of approaching a cliff.

The flying lizard algorithm
Adam implements an unsupervised machine learning algorithm affectionately known as the pterodactyl. This software spots surprising similarities among cancelled accounts. Later Adam was able to train the great flying lizard to identify accounts in imminent danger of cancellation. This information empowers the marketing and customer success teams to take proactive action. Uncovering these surprising similarities gives Adam’s team actionable ideas for cost-effective retention strategies.

Analytics Automation
Systems Supervised Machine Learning Intelligent Automation
Innovation Unsupervised Machine Learning Expert Systems
Management People Analytics Supervised RPA
Operations Process Analytics Unsupervised RPA
Resources Natural Language Processing Learning Bots

What's Your Story?


71-75 Shelton Street, Covent Garden
London, WC2H 9JQ
United Kingdom