Over the last couple of years, how many times have you wondered if your phone was actively ‘Listening’ into your conversations? How many times did you get an Ad on the browser, promotional email or a video as a recommendation that was related to a conversation you had with a colleague, friend or partner? Making you quite paranoid about your privacy? What if I tell you that these recommendations are from recommendation engines that are powered by state-of-the-art data science? ? More details in the article, welcome to the age of Hyper personalization.
Imagine a local storekeeper that is friendly, curious and maintains A good relationship with all his customers. The business would likely thrive as the storekeeper would be able to anticipate the demand for certain kind of products and would stock-up accordingly. This is what used to happen a lot in the pre-industrialization era. There was a certain degree of personalization right from the greeting to creating a custom experience. But this was not a model that could scale, at that point in time.
The dawn of industrialization increased the production capacity, the supply increases, but there is a lack of personalization. You may recall Mr. Henry Ford’s famous saying, “My cars can come in any color, as long as they are black”. Businesses had to sacrifice personalization to achieve scale.
This was followed by the birth of market research. Businesses started understanding their customers better, user segments were created and each customer was mapped to a segment. Product recommendations were made based on the segment a customer belonged to. This allowed a better level of personalization and at a scale that was never seen before.
Then came the GPUs. The dip in the cost of computation and increased affordability of storage allowed companies to start looking at data as a valuable asset. Soon data took over oil as the most valuable asset on earth. The platform model of business allowed more data to be collected, curated and shared resulting in more usable data. The segment sizes started shrinking and today most of the tech giants have a segment size of 1! They can curate and cater to each individual better than the smart storekeeper of yonder days.
This is all a known story, but how did companies get there? How are they collecting data and how they can recommend content from a conversation I had with my colleague? Is this possible? To dig deeper into this, we need to take a detour into understanding human psychology — Illusory correlations. It is similar to the thought process where your waiting queue always seems to move slower, rather you register it strong only when it moves slower.
Similarly, you tend to notice only the content that had some correlation to a previous conversation or thought while the rest of the recommended content fails to register. These recommendations are not mere coincidences — by using the geolocation of people close to you and their consumption patterns, algorithms determine what you could be interested in. If you take a look at your Youtube, Amazon or Google feeds you will see this pattern emerging yourself.
Businesses are using the data they collect in interesting ways. Any business that doesn’t provide hyper-personalization will fall behind the competition. Once the process of creating and using data is demystified, it becomes easy to carry out. This involves innovating at the data collection layer in UX to using Machine Learning models to create predictions for individuals.
Written by Phani Ranga, VP Digital Consulting