The Crucial Role of Quality Engineering in Hyper-Personalization
Hyper-Personalization is changing the way we enjoy our leisure time. It is more advanced than traditional segmentation techniques and thus presents unique challenges to companies to ensure quality.
No one knows how long personalization has been an integral part of our society. There are many ancient examples, from cave art to initials embroidered in a businessperson’s cuffs. It is no surprise then that we are using technology to create an experience like no other for our customers. This trend is known as Hyper-Personalization, and it is changing the way we enjoy our leisure time.
At its core, most would agree that true Hyper-Personalization creates an experience that is unique for every user, even in systems of millions of users. It is more advanced than traditional segmentation techniques and thus presents unique challenges to companies to ensure quality.
AI is key to Hyper-Personalization
The experience produced by Hyper-Personalization dictates we use AI (Artificial Intelligence) and ML (Machine Learning) to be more than a segmentation personalization system. For Product Owners, this creates a unique challenge for we can no longer test complex systems with the expectation that if you input a value, you can predict the output. Think of a recommendation engine in a movie and TV streaming service that we want to create a custom experience. With millions of shows we could end up with a near infinite number of personalization/recommendation options for our users. How do we test a system like this?
At Qualitest we have been investing heavily in AI to answer challenges like our movie example. To ensure people’s enjoyment of platforms requires innovative approaches to decades old testing where we must consider technology, society, and the final outcomes. This means a mixture of agile test planning mixed with innovative technology and in many cases final acceptance done by a crowd testing approach.
What makes up Hyper-Personalization?
Here are several common inputs used by personalization engines when we go beyond pure market segmentation models:
- Big Data for our users
- AI & ML
- GPS data of user
- Time of day/year
- Regional data
Take our earlier example about a movie service. If we knew our user was a child, it was December, and they are in the US, we might recommend “Home Alone” to them. Testing these AI systems means one must utilize new techniques that can build known profiles and have more right, less right, and in some cases wrong tests. A movie rated ‘R’ would always be a wrong choice for a youth.
What about a store that has overstocked surf boards in the summer? They might want to use geo-tagging to notify all users within 30 miles about a 30% discount via a social media platform frequented by surfers.
So, what do we need to go about ensuring quality with all these systems? There are several requirements we find most often with our clients.
- Target Personas and Test Models for Hyper-Personalization
- Automated Mobile Testing Platform for Geo-Requirements
- Crowd Testing
- Legal Testing
Hyper-Personalization is here to stay and has become a mainstay to engage your customer base to maximize brand loyalty and lifetime customer value. The testing procedures required for testing these new systems is very different from traditional QA. One must use new models to test correctness in ranges versus absolutes while still checking for wrong answers.