Relevance V.S. Novelty
Why YouTube intelligent has driven me nuts
No matter it is a web 1.0 or web 2.0 platform, with time going on, it becomes heavier, rich content becomes their treasure but also brings challenge. The challenge is, how to present to the customers with the right content they really want?
With machine learning and big data, the system has become more and more intelligent and the platform wants to really understand customers and present to them what they will be interested. Of course, this is not an easy job, even a couple who have been married for many years may not exactly know what the other one really wants, how can a system achieve that? What the machine has done was simply to analyze the data collected, identify the patterns, then make decisions for the customers. But humans are not that easy to be predicted, and such intelligence may not receive much appreciation.
1. Customer preference is not a white or black thing
By default, people tend to think that if one like one item/thing, then he/she will not like other dissimilar items/things. For example, one of my female friends asked me about my opinion, because she had a crush on one of my male friends. But she was so worried that he would not like her because his ex was a totally different type of girl. That story was a happy ending, those two friends of mine end up into marriage with two kids now. So, it is not that, if one likes type A, then he/she will refuse type B. Customer preference is not white or black thing, it can be gray, or it can be both.
If a real estate agent decides not to show his/her client contemporary style of homes only because the client showed interest in a Mediterranean style previously, he/she might just lose the opportunity. A person who often spend vacation at countryside, does not mean he/she dislike the city life. Choose one does not mean against the other one. Even though in many cases, customers do have a tendency, but if the options have been removed, only because the system thinks that it has found the user pattern, customers will not happy about that.
2. Understand triggers and craft customer journey
Machine learning is rationally driven in data analysis while customers' purchasing behaviors are actually emotionally driven. People tend to believe that choices were made out from rational analysis of alternatives. However, in reality, that is not how they behave. Emotion actually might be the determining factor for most purchasing decisions.
I am not denying the importance of machine learning and presenting to the customers with relevant displays. But I do think besides that relevance, we should also provide something interesting, something fresh, or even something unexpected. In order to generate more sales, we need to drive customers’ emotional react, which means, we should provide customers with “emotional motivators”, which will trigger their emotional reaction, hence make a purchase decision.
To design the emotional motivators, we should understand the triggers and craft customer journeys accordingly. For example, customers might want to make a purchase when they think they are picking up a bargain, not because they really want the item, but they’d like to take full advantage of it; customers may make purchase because of their fear of missing out. They do not want to miss current or future opportunities, including financial opportunities, rewarding information, or a new trend in fashion; customers may want to buy something because of jealousy, pride, or vanity, hence make a purchase of something they may even hardly can afford; customers may want to buy something which makes them feel belonged to. It is ashamed to be left behind by peers, hence they might just buy something to match others; some people may make a purchase for sentimental reasons, which raises their good memories, or beautiful dreams, or lifestyle pursuit.
3. Why I hate YouTube intelligence
Even if it was your mom or your spouse, who thinks she/he understands you, then she/he just made the decision for you, and always bring in the same choices of food or clothing, and assume that is for your good, you will be mad, right? That is exactly what YouTube intelligence has done.
Only because I watched some “Three Kingdom” movie, then the whole screen was almost full of Three Kingdom series. Even those I have watched hundreds of times, will keep coming up again and again. Gosh, give me a break! Bring me something new, please! YouTube intelligence does not make me feel fun, instead, it makes me feel trapped. It makes me so hard to find something fresh, make me so hard to keep up with the trend, and I even hard to know what is popular today. Every time I log in to YouTube, it is just like my own collection of all the old items, instead of a place for me to keep up with the world, or a place for me to experience new experiences.
4. If customers rely too much on keyword search, instead of browse, that means fail
Many e-commerce platforms may have noticed that their customers tend to use keyword search a lot more than browse, and they conclude this as people’s needs for efficiency. I do NOT agree.
Per my research, there are many customers like to spend a lot of time browsing merchandises on e-commerce platforms, even though he/she may not have something in mind that was planned to purchase. Even for those who are busy at other things, and do not spend so much time browsing products, he/she would not refuse to dig more if the platform really has something caught his/her attention.
Browsing is like window shopping, it is part of the pleasure of shopping; such joy will be greatly decreased if only search and find the right item then make the purchase in a short period of time. Such efficiency is the killer for the happiness of shopping. Just think about people go to the mall on a weekend with the family, do they just directly make the purchase of the items needed and then leave? No! Wandering around, waiting in line to taste some ice creams, try this and that, eat at the crowd food court, let kids play at the small game station… All these are part of the fun of shopping. How many percentages of items purchased actually was not as planned? Browsing actually will generate a lot of much more emotional sales than rational search and find sales.
5. Repeated pattern would not increase customer satisfaction, novelty do!
Some platforms do not have customized homepages, almost everyone come to the site will see the same content, of course, that is not an ideal way. Some platforms like YouTube have used machine learning, and present to different customers with personalized pages, which could have been a good thing, but we should never forget to stay novelty as well, besides relevance.
Neurologically, novelty will increase the release of dopamine in the brain. Which means, when people see something new or unexpected, they feel excited and more focused. The curiosity will drive people to dig more and dive deeper. In Wikipedia, it described “novelty seeking is a personality trait associated with exploratory activity in response to novel stimulation, impulsive decision making, extravagance in approach to reward cues, quick loss of temper, and avoidance of frustration.”
If machine learning is to understand user pattern and provide customers with something within their taste; then novelty is to bring customers with something new, make them happy through unexpected discoveries by accident. So, besides providing something relevant, based on customer understanding; we should also unleash the unexpected, and provide customers with something fresh and interesting. We need to find a good balance between relevance and novelty!
6. Make machine learning smarter and fun
I am sure the machine learning algorism have been improving. Besides relevance, accuracy, novelty will also be taken into consideration. Besides human’s behavior pattern, human’s emotional pattern and expression recognition will also be combined into the algorism. Skills such as persuasion, empathy, humor and understanding may also become part of the robotic skills. When technology has been developed to that level, I am sure, people won’t have the problem I encountered today with YouTube.
As we know, machine learning goes beyond Big Data analytics, which means, it is not just simply collect and statistic the data, but learn through user behavior and then autonomously adapt. With more advanced algorithms employed, and the machine becomes smarter, with EI (emotional intelligence) get developed, the system should be able to emulate the thought process behind human decision-making; and the system should also be able to provide customers with not just accurate, relevant results, but also novelty recommendations.