Traditionally, marketers would try to segment their customers into groups, based on characteristics such as demographic information, purchase history, browsing behaviour or geography. Once customers were placed into a segment they could be targeted using rules defined by the marketing team. While this approach is better than marketing without any targeting at all, is necessarily broad and unable to deliver a genuinely personal experience – if you have millions of customers, you are likely to have many thousands of customers in each segment being marketed to in the same way based on the same rules pattern.
Amazon.com has proven that machine learning (ML) is a great way to create a ‘segment of one’ – using ML to write specific rule sets for each customer that deliver them an optimum experience based on their unique characteristics, and continually update over time as the customer, the market and their business changes. While Amazon.com has been using ML in this way for over a decade, it’s been difficult to do for many businesses. Customers tell us a lot about their preferences with their behaviour on our digital platforms, but these behavioural data sets tend to be large, change fast and are generated in spiky patterns. Just collecting and analysing this data was difficult and expensive for customers until services like Amazon Kinesis were released to simplify the data collection process.
With many customers now having the data they need on their AWS account, customers wanting to use ML to generate genuinely personalised experiences still need to select and build ML models, train, deploy, monitor and manage them, and then set up a presentation layer (usually an API) – a lot of heavy lifting! Luckily AWS has released Amazon Personalize which takes away all of this and does it for you, comfortably scaling up to handle the requirements of the busiest digital platforms.
While we have delivered some great personalisation outcomes before, Amazon Personalize dramatically reduces the effort required to obtain those outcomes. Our consultants at Intellify spent a week with the Amazon Personalize team at AWS going through best practice when implementing the service. We’ve brought it back to roll out in our personalisation projects and it’s enabled us to get great recommendations generated in a fraction of the time and effort it required before.
If you need help with your personalisation project, contact us today!