The idea of an “online assistant” is generally not new. However, there is a lack of good practical implementation in E-Commerce. Startups and corporates are currently working on different parts of the user journey with multiple emerging technologies to improve existing KPIs in Cross & Upsell and in the reduction of return rates.
But what are the needs of the user in online shopping?
There are two main different intentions of the user entering an online shop:
1) The intention to find and order comfortably a concrete product that user`s have in mind from a specific brand. In this case, the user follows the general filter possibilities to get fast to the desired product.
2) The intention of an inspirational journey through an online shop to get inspired by products in a specific product category or to browse generally for inspiration. If the user comes across products that he likes, a purchase should follow.
In both cases, finding desired products can be a challenge if the online shop has tons of (similar) products. A passive recommendation (“others bought also this…”) won`t help in this early part of the user journey.
In comparison to a shopping experience offline in a physical store, user can`t try on the products online. They can`t see or touch the products to see if
· the size fits to the individual body form and whether
· the colors and / or the material also correspond in reality to their own taste
These are the main reasons why user either order a lot of products / sizes and return the majority afterwards. Another scenario is that user fill the basket with a lot of products and leave the online shops afterwards without a purchase, because of their insecurity in the above mentioned points. If the online shop has also a physical store, the probability is high that the user will go in the store first to try on the products and purchase them directly offline, if there is no price difference.
What are the current known solutions and how does the user perceive them?
The most known and implemented solution to give support during the online shopping journey are Chatbots. They should help to navigate the user easily to the desired products – 24/7.
Markets for Chatbots grow internationally: from USD 2.9 billion in 2020 to USD 10.5 billion by 2026, at a Compound Annual Growth Rate (CAGR) of 23.5% during the forecast period. Among verticals, retail and eCommerce vertical segment is expected to hold largest market size during the forecast period.*
Covid gave this growth a bigger boost. These main following needs increased during Covid:
· Need for 24/7 customer support at a lower operational cost
· Rising customer demand for self-service operations, which led to a focus on customer engagement through various channels and advancements in technology
In a study from Zurich University of Applied Sciences in 2021 about Chatbots, user also claimed the 24/7 availability and the speed of interaction as the No. 1 advantage of chatbots. Especially for users browsing in the evenings or on the weekends.
Nevertheless, chatbots still need a lot of improvements. Current deployments of chatbots can be found mostly in customer service areas and process improvements such as delivery status tracking. Chatbots used in a user shopping journey ask the user about their clothing style tastes to assure a fitting product recommendation. This takes a few iterations and more important: no “real” conversation can be found in those solutions.
When it comes to user interaction, chatbots highly lack of personalization. Especially when it comes to a potential personal assistance when choosing products like clothes or shoes online.
User`s associate the chatbot with a machine and not with a personal assistant. Thus, the user knows that the chatbot will not be able to help understand the fitting and sizing of a specific product.
Conversational AI as a possible solution for personalization in Chatbots
The current main goal in the market is to improve the chatbot`s human interaction and therefore also the user experience. In July 22, Forbes published an article about Conversational AI where the user`s needs were described like:
“When I need help, the app should “understand” me, know exactly what I do, what I mean, what I need and what my perfect outcome should be (in multiple languages in ultra real time). Is that too much to ask? Today it is. But that’s where this needs to go – and not just for customer service, but for all interaction with intelligent systems.” **
Self-learning, behavioural driven data chatbots can adapt to changing conditions in the environment they operate in and can learn from their actions, experiences, and decisions. These chatbots can be considered intelligent enough to enable the delivery of more human-like and natural communication.
But when it comes to verticals, like e.g. shopping clothing and shoes, which additionally requires industry-specific requirements, it still becomes a challenge. As described very well in the Forbe`s article: current chatbot solutions fail in human interaction and voice solutions like Alexa are too generic to solve industry specific user needs.
The term and degree of “personalization” needs also to be defined in that context. Conversational AI is built upon past behavioural communication process data. Marketing Experts know that behavioural data is absolutely necessary, but are not enough for effective communication. In many years of empirical research, Brandmind could prove together with Zurich University of Applied Sciences that psychological types communicate differently and prefer different sensoric elements (material, colors etc.). The adjustment of the communication style (e.g. using different words to express the same topic) and product recommendation within a conversational AI for effective communication has not been taken into account until now.
Virtual shopping assistant – the online sales person
A recent user research study, conducted by the Zurich University of Applied Sciences and Brandmind, revealed that the expectations towards a virtual shopping assistant would be to help the user to
· get faster to the desired products
· give more detailed indications about fitting & sizing, the material and the colors so that the user feels more secure in selecting the “right” products
· give the user the feeling that the conversation is being done with a person and not with a machine
Thus, all the above mentioned challenges within Chatbots and Conversational AI should be solved with a virtual shopping assistant. Additionally, the appearance of a Chatbot should also get a human touch and should be placed in different, but concrete areas of the Website to not disturb and create value within the user journey.
How is it possible to achieve that ambitious goal?
The key to success is a mix of existing data (product information, behavioural data) and an additional datapoint like psychological typologies and their communication style as well as their sensory preferences (material, colors).
Startups and Corporates often work only with “existing data” and train models with Machine Learning. But as it is in classical Marketing: existing data isn`t enough to perform outstanding communication and sales. When it comes to personalized communication and sales, human design knowledge management needs to be the basis for a self-learning AI solution. Machine Learning is only used to train the pre-defined knowledge in big data.
Customers of E-Commerce Shops should be anonymously psychometrically profiled before the virtual assistant goes online so that the approach towards a customer once logged in, is different. The information about purchase history and their personality profile leads to high customer experience online. Unknown users undertake a few seconds entertaining game to assess the personality once so that the communication style and the product recommendations can be adjusted towards the personality type. The user will be guided until the process of the purchase starts to assure that the user feels confident in all the decisions.
Brandmind received funding from Innosuisse (the swiss state) to develop a virtual shopping assistant together with the Zurich University of Applied Sciences, the collaboration partner Coteries and with a swiss implementation partner in the market. The goal is to increase Cross & Upsell KPIs, User Satisfaction and reduction of return rates. Beginning of 23 a first MVP will be ready to showcase and sell.
Learnings to meet better User`s needs
Data is King. This is also not new. Despite the fact that many companies still need to improve massively their internal Data Management, the most important aspect is to learn which kind of data is relevant to meet user`s / customer`s needs. The exact same mechanism as in classical Marketing, even though it is still generally assumed unfortunately that existing data is sufficient to deliver an excellent customer / user approach.
The ideal mix of data to provide outstanding customer approach or user experience are
· customer behavioural data (product preferences in the past)
· socio-demographic data AND
· insights about their psychological type (communication style) and the according sensory preferences (material, color etc.) which help to navigate better product tastes.
Especially the last data aspect of psychological typology is new to the market – for marketing in general, but also in AI solutions. Adding an additional and very important data point is crucial for an effective online user & customer approach. This additional data point should be matched with existing data, because it helps to improve significantly the marketing mix, product recommendations and finally also the desired customer behaviour leading to a purchase.