The future of psychometric recommender systems
BrandMind’s Approach to AI-based models
Around a decade ago, people were shopping in a physical store where the number of products was limited. Due to the emergence of big data and an increase in the store size, users have been provided with a lot more options than before, hence it is more important to ensure the user is recommended the products that align with his interests, or it might lead to a drop in the company’s profit value. Due to increased options and fewer resources, the manual product recommendations are now automated using AI to build a recommender system.
Retailers like Amazon or eBay use AI-based recommendation models to power their user-click behavior thus generating more revenue and focusing on producing the most demanded products. Figure 1 depicts the classic approach to revenue generation through user interaction, while Figure 2 depicts the revenue generation through user interaction with our model.
The approach that BrandMind takes can be visually understood from Figure 2 — when the user enters an e-commerce platform, he starts interacting with the platform by searching for products and clicking on products that are the most appealing. During the interaction, our models profile the user based on their psychometric personality. Every user is then segmented in each of the profiled personalities and the products based on the personality appeal are recommended to the user. Since most of the web-search engines and e-commerce platforms use the taxonomy approach with collaborative filtering, we intend to emphasize those two aspects in our AI models along with a key factor of psychometric personalities.
Recommender systems are bifurcated into five types:
- content-based, the user is recommended items that are content-similar to the items the user already liked, we consider the user history in this case;
- collaborative-filtering, the user is recommended items that people with similar tastes and preferences liked in the past, we consider current trends and the user history of all users to then form association;
- hybrid, the user is recommended items based on a combination of both collaborative and content-based methods;
- taxonomy, given an existing taxonomy and a set of new emerging concepts, the aim is to automate the expansion of the taxonomy [1] to incorporate the new concepts without changing the existing relations;
- psychology-based, using the concepts of psychology to recommend products to the users in the same sets.
Apart from the aforementioned techniques in recommendation systems, we introduce a key factor, that as per our observations has reformed the user-click history and has led to exceptional conversion rates, which is the psychometric profiling of the user — what I mean by that is that the user will be a mix of multiple personalities and we retain the gray areas rather than segmenting users in specific classes. The blog series is going to comprise a few other blogs where I walk you through more insights on the work and approaches of BrandMind and as we proceed it will also provide an idea of our AI model.
- Why do we target psychology-based recommender systems?
Psychometrics focus on the user’s implicit behavior rather than the explicit ones, to understand the difference between the two let’s take two examples.
Scenario one is that you are actually pretty upset and your friend texts you. In this case, your friend wouldn’t know that you’re upset unless either of you explicitly discusses it, this is how explicit behavior is tracked — by gaining more information from the customer.
Scenario two is that you are still pretty upset but your friend now comes to meet you suddenly. Now in this case opposed to the other one, there is a high chance that your friend would notice that you are not all right even if you do not mention it.
This is the way that implicit behavior works and research has proven that unconscious emotions are the main triggers for every action that we take. In order to track these unconscious actions, we track the implicit behavior of the user. The current market is more focused on the explicit triggers to understand the user behavior where direct questions are being asked but asking direct questions prevents from entering the unconscious behavior of the user. Whereas, for our methods, we use implicit behavior to gain indirect information about someone by engaging them with pictures to get into their unconscious emotions. - How are we so confident that it works?
Since BrandMind also works in consultancy projects, we have tried and tested our methods on the ground level with real customers and users, have provided companies with a holistic overview of their customer segments, so they focus on the products that actually bring them the most profits and are in demand based on the highest number of active customer types (also illustrated in Figure 2).
Using our existing packages with the ongoing clients and customers, we have seen our clients expand their profit margins based on our research and service. Normally in direct marketing, the rule of response is 1–3% whereas, with our services, we brought it up to 11%. Another one of our success stories was when one of our clients took another marketing campaign for power-oriented people and our sales escalated from 0% to 3% — one of the key reasons why companies would want to get involved. - What did we exactly do to understand the customers?
The ongoing work is a comprehensive package offer where the basic end includes consulting by simply relying on psychometric knowledge all the way through to more expensive packages where we can run some neuromarketing techniques including EEGs or MEGs, etc. Customer profiles on a larger basis though will mostly be completed by a survey questionnaire of images and their associations which will be sent to their clients. Once the customers are profiled we offer to establish the best marketing methods based on what product or service appeals more to the personality type of a customer thus generating greater interaction.
References:
[1] J. Shen, Z. Shen, C. Xiong, C. Wang, K. Wang, J. Han, TaxoExpan: Self-supervised Taxonomy Expansion with Position-Enhanced Graph Neural Network (2020), WWW ’20: Proceedings of The Web Conference 2020
[2] J. Shang, Psychometrics: How Consumer Attitudes Influence Buying Behavior (2020), The Adroll Blog
About the Company:
BrandMind is a consultancy turning into a Marketing-Tech company that creates emotional experiences across all (digital) touchpoints. At BrandMind, everything is about triggering the emotions of consumers and providing an outstanding customer experience based on scientific know-how. We specialize in the combination of neuromarketing, psychology, and artificial intelligence. With in-depth knowledge from brain research and psychology, we can apply our know-how to all touchpoints and measure it in the future. Due to our scientifically founded tool, we can provide psychometric marketing for companies in the future with which the relevant Online-KPIs, such as (CR, Sales, etc.), can be massively increased.
