Here are three examples from B2C and B2B e-commerce that illustrate how retailers, distributors and manufacturers can combine human intelligence, machine learning and data to improve results.


Mihir Kittur, chief commercial officer, Ugam

Rewind to the time when the main factor of production in the agrarian economy were farmers. As we moved into the industrial age, human expertise remained crucial, but machines began to play a greater role. Then came the age where the primary factors of production were computing power and data. Today, we find ourselves in yet another chapter. Some may call it the “age of the customer,” while others may call it the “age of decision automation.” But what you call it doesn’t matter! Unlike the earlier waves, where the emphasis was on only one factor, today there is an interplay between all three factors: human expertise, intelligent machines and data. And that development is disrupting and accelerating several industries.

Retail is no exception. Most retailers are using data and analytics to unlock opportunities and automate decisions in real time; however, this rapidly evolving environment is also a cause of anxiety for many. Some have invested in big data infrastructure, but question whether it is driving meaningful impact for their business. Others wonder where to start and fear they are missing a major opportunity. Here are some examples and lessons learned from large retailers, distributors and manufacturers who leveraged data and analytics to maximize impact.

With a thriving brick-and-mortar business, a leading mass-merchant retailer wanted to become e-commerce-ready to cater to the growing community of customers who begin their shopping journey online. The company knew it needed to onboard SKUs online quickly, but it had limited product data and lacked product content standards. To add to the complication, 60 percent of its products were private label and so it was difficult to acquire credible product information using traditional means.

First, ask yourself if the problem is worth solving.

Machine learning held the answer. It enabled the extraction of attributes from various sources, such as packaging material, and classification of the resulting information at scale. Then, using natural language generation, it automatically converted the acquired information into product descriptions. This combination of client expertise (which was crucial in determining alternate sources of data) and sophisticated tools and data enabled the retailer to onboard SKUs in one-third the time it would have taken otherwise.


Next is the experience of a large B2B janitorial supplies distributor. Following a major acquisition, the distributor found itself offering such a wide-ranging product assortment that it was eating into the company’s margins. Hence, the company needed to rationalize its assortment. Unfortunately, the distributor didn’t do an effective job of letting its customers know about alternate products for those that had been eliminated. Customers didn’t always find what they wanted, and for 80 percent of the out-of-stock products there were no recommended alternates. This led to poor Net Promoter Score and a 10 percent drop in sales.

To find a solution, product category experts worked with analysts to help them understand each category. After obtaining, reviewing and blending data, various matching algorithms were constructed and fine-tuned, based on the category managers’ feedback. This human plus machine approach increased product coverage of alternate products from 19 percent to 78 percent, improved customer experience and resulted in a five percent increase in sales.

For a global luxury manufacturer, the challenge was different. The company didn’t have enough visibility of its brands online. The company’s leaders wanted to track key performance indicators such as share of shelf, out-of-stock levels, pricing and promotion compared to competitive brands, etc. They also wanted to know if their SKUs were priced low enough to be competitive or, alternatively, if they were too low, which could impact brand’s image. To acquire such competitor data, the manufacturer turned to sophisticated big data technology to capture information from different websites in different languages and analyze the information gathered to derive insights. Combining that intelligence with human expertise, they were able to help drive better data-driven decisions across a whole host of KPIs.

Each of these use cases yielded lessons that can be applied to effectively use data and analytics for maximum impact.

1. Ask yourself if the problem is worth solving: It is important to invest time in framing the problem and understanding if it is worth solving. Be careful to avoid putting the answer being sought before the problem.


2. Stay committed and invested in solving the problem: Once you’ve defined the problem, it is important to commit yourself to solving it even if you don’t succeed at the start. Be prepared to try, fail and iterate before achieving the outcome you need.

3. Accept from the start that you can’t do it all alone: Digital disruption is taking place so quickly that few companies have all the skill sets required for success at any given point in time. Find partners who have a track record of success and will substitute for the skills your team lacks.

4. Realize that clean data requires hard work: To glean insights from your data, it is imperative for data to be clean and complete. Maintaining such a database requires constant diligence and discipline. It’s very important to get this right, so seek help if it’s needed.

5. Improve the organization’s literacy around algorithms: It is beneficial for those involved in data and analytics projects to understand the basic terms associated with algorithms, how they work and their limitations. Today, there is ample information available, so ensure that you are, at the very least, aware of the most relevant and basic information.

6. Create a “learning ecosystem”: Large-scale data and analytics projects are not “check box” endeavors and are most effective when lessons learned are continually reapplied to additional projects.


Back to the farmers: The human expertise and the tractor that enabled them to dramatically increase their output remain as important as ever. But today’s large-scale growers also tap into modern technologies, such as satellite-powered global positioning units to guide those tractors as they plow the fields, or drones to survey the health of their crops. As a result, a farmer today can yield far more crops than what was possible even a few years ago. Similarly, in retail, those who combine human expertise, intelligent machines and data to solve the right problems will be the ones who experience transformative growth.

Ugam provides data and analytics for retailers, brands, B2B manufacturers and market research firms.