3 key advantages for AI in the retail space
Source – venturebeat.com
Accenture suggests that the core of retail strategy is a 720-degree view of customers — reaching digital natives with rapid focus shift, high expectations, and growing demand for personalization and perks. Successful examples such as Amazon, Macy’s, and Walmart prove that the way to reflect and interpret this view goes through experiment and innovation. In particular, the use of AI and its integrals includes data mining, machine learning, natural language processing (NLP), and bots. But how is AI a good match for retail? Here are three advantages.
1. AI can extract value from massive data sets
Industries have been struggling to build data-driven strategies for a while. According to the numbers from the last year McKinsey research put out, retailers were second in this race. Retailers are lucky enough to collect and own massive data about customers and buyer behavior. However, they have been unable to transcript this data properly.
According to the research, retailers extract value only from 30-40 percent of existing data. That leaves two thirds of data wasted due to a lack of process, technology, and analytical talent. Besides, most of the data stays “siloed within companies.”
Improvements in machine learning and, most importantly, data availability help retailers unlock the full potential of customer data. On one hand, a regression model allows a retailer to leverage legacy data and reuse it effectively. On the other hand, predictive capabilities of machine learning let retailers not only learn from experience, but also apply those insights in order to model and predict future buyer behavior. It’s a true virtue to know what customers want before they want it.
Example: Walmart uses machine learning to predict what its ecommerce shoppers are likely to buy. Thus, the retailer provides focused recommendations based on past customer behavior. This is what the VP of customer experience engineering at WalmartLabs calls “the bridge to enhance online shopping experience.”
2. Customers need to be understood
The market of in-messenger and voice assistants is growing. The results of the 2017 Prime Day sale let Amazon claim Echo Dot as its best-selling product.
At last, people gain the power to speak with digital systems using natural language thanks to chatbots — in particular, the bots enhanced by NLP engines by Google, Amazon, Microsoft, IBM, Facebook, and soon also Apple. And people like it.
NLP engineers, in turn, finally get access to real-time data in natural language pumping from messengers, web, and voice assistants. This data is the key to maturing AI, and therefore to truly intelligent and helpful systems.
Meanwhile, retailers, especially big ones — which remain the driving force of NLP development — already use the opportunity to understand their customers. They sell and upsell via AI-driven chatbots and Alexa skills.
Growing attention and use cases feed both business and technology, proportionally. The more customer queries are parsed, the better NLP systems understand natural language. The better NLP engines work, the more customer needs are met and products are sold.
Example: The range of Amazon Alexa skills is already pretty impressive, from recommendations on books (Pan Macmillan), wine (MySomm), and music (Spotify) to full automation of cab ordering (Uber), pizza delivery (Domino’s), and household services (Laundrapp). These skills literally sell using voice.
If voice assistants don’t do direct sales, they at least make them smooth. Macy’s On Call, based on IBM’s Watson engine, enhances customer experience and helps shoppers find items in a retail store with no human involved.
3. NLP and machine learning provide real value
We’ve learned how to understand real-time customer queries via NLP and extract value from legacy data using machine learning methodology. The challenge of making use of ongoing customer feedback is bigger, but so are its benefits.
This challenge requires joint forces. First, an NLP engine needs to extract sense from a query in natural language. After, machine learning steps in to extract value from this sense.
Using classification, intelligent machines assign meaning to data, relying on their background and existing knowledge.
In practice, the system classifies certain products, say “books,” by categories, say “popular among women over 65.” For retail, this means more focused recommendation and upselling.
Using clustering for new information, in turn, opens totally new horizons. This method allows the system to find patterns and build connections between the bits of information with no set criteria and thus without prejudice.
In practice, it means that machine can find unlabeled, non-standard connections between customer buying habits. It can understand why people who have read X books by author A will most likely to go for a book by author B despite a whole range of other authors in their category. For retail it means more than a focused recommendation. It means more intuitive recommendation, better service, and higher customer satisfaction in the long run.
Machine intelligence market has reached more than a hundred billion dollars in value and keeps growing. It doesn’t seem likely to fade anytime soon, either, since all the cool kids are in it (Google, Amazon, Apple, and the like). What’s more, AI giants are striving to make it both available and affordable.
In this context, the retail industry doesn’t have much choice but to embrace AI. Directly connected to and dependent on customers and data, retailers start using AI as an experiment. But soon, application of intelligent machines will become a competitive advantage. Then, it will turn into a necessity and a part of every retailer’s business strategy.