Abhishek Kumar - 10 months ago

In marketing research, a lot of research is done to provide insights into how at psychological and market level customers behaviour works. Data Scientists from marketing research domain have been taking about ideas especially, needs, wants and demands in content of recommendation engines. 

The understanding needs & wants models to provide a wide angle perspective   of Here I just want to give a refresher on types of need with the popular car example. While reading through the example try to think what needs our recommendations engine fulfill by churning volumes of data.

There are five types of needs as per marketing research.

  1. Stated needs - The customer wants an inexpensive car
  2. Real needs - The customer wants a car whose operating cost, not initial price, is low
  3. Unstated needs - The customer excepts good service from the dealer)
  4. Delight needs - The customer would like the dealer to include an onboard GPS navigation system
  5. Secret needs - The customer wants friends to see him/her as a savvy customer

Now, as a Data Scientist I need to ask, which needs are we capturing in recommendation systems and which needs does actually recommendation system fulfils by predicting?

The recommendations engine is built on the purchase history. The recommendation becomes strong by each positive feedback you give to the system. So, if you started browsing shoes of any kind, the system builds a hypothesis that you like shoes. Now every moment you click on shoes or buy, the signal gets stronger and your choices get narrowed by the recommendation engine to shoes. These are psychological cages being created by engines which drift customers away from variety and fulfilling needs except stated needs.

Over period this can eventually cause permanent changes in customer behaviour and market structure;

  1. Monopolies of brands and tough entry barrier for alternative products which can serve similar needs for someone. e.g., I may buy shoes for gifts, for running, for hiking. But my needs are -> gifts - making someone happy, running- fitness, hiking -adventure.
  2. In the online marketplace, few items and brands can be cornered and eventually killed due to recommendation engines positive feedback leaving them out.
  3. They can impair the natural human ability to try new things & learn by experimentation. A deeper personalisation creates a spiral to converge at a point where you become a product.

Data Science is going to change a lot in our daily lives and will give rise to a new civilisation. We have seen these transformations happening in Industrial revolution, Computer revolution and now witnessing in AI revolution. It is important for us to frame new era of AI using our collective knowledge of centuries to create a sustainable and in tandem with nature civilization.

Ethical use of data science for betterment has been at core of sustainable development.


So, what are next steps?

Probyto can work with you to understand how your recommendation engine is working and influencing your users. Limit the adverse effects of recommendation engine and keep happy growing customer base.

Contact Us