E-Commerce Runs On Data Powered By AI

For e-Commerce businesses, AI adoption is no longer optional — it’s a necessary step toward streamlined operations and market competitiveness. And while it’s clear intelligent systems are a necessity, there’s little guidance on how to overcome foundational barriers like sourcing and annotating data at scale or developing a practical AI and training data strategy.

AI Adoption In Retail

There are varying degrees of AI adoption across industry sectors, with telecom, high-tech and financial services leading the way in overall AI adoption, while retail is excelling in AI for marketing and sales, as well as supply-chain management.

In fact, according to a McKinsey Global Survey, 52% of respondents representing the retail industry reported using AI for marketing and sales, compared to an average adoption rate of 27% across other industries.

In the same respect, 38% of respondents representing the retail industry reported using AI for supply-chain management, compared to an average adoption rate of 16% in other industries.


This preference for deployment of AI comes as no surprise since it arguably creates the most value for the retail industry — i.e. a better understanding of customer behavior and streamlined end-to-end fulfillment of orders leads to increased revenue potential.

Mapping AI Technology to Business Problems

McKinsey and Company reports that 47% of organizations surveyed have at least one AI capability embedded in their business process. Approximately 30% were still piloting AI, and the remaining companies reported using AI as a core part of their business.

It can be challenging to assess AI opportunities in a way that ensures a greater overall value from using AI when compared to the overall investment. To better understand how AI technology maps back to business problems, consider AI adoption across three categories: sensing, predicting and automation.

Using AI to sense can range from detecting emotions and intent, to detecting anomalies in performance, availability or quality. This application of AI can help improve quality assurance processes, consumer segmentation, retail item cataloging, etc.

For example, among Walmart’s machine learning initiatives, one use case is to ensure its retail inventory is comprehensively cataloged. If an item isn’t cataloged, that may impact reporting of inventory levels and misrepresent availability of the product to the end customer.

In this scenario, AI technology is used to classify products so Walmart systems can accurately sense availability. With help from a trusted training data annotation partner, Walmart was able to use machine learning to improve its retail item coverage from 91% to 98%.

AI technology that can be used to make predictions in a retail setting can range from demand forecasting to optimizing customer experience based on previous buying behavior. Amazon’s robust product recommendation engine was reported responsible for 35% of the company’s revenue. However, despite the boom in e-commerce sales, brick-and-mortar retailers are still expected to account for 85% of U.S. retail sales by 2025. This makes predicting scenarios like how weather forecasts influence store traffic just as critical as recommending the next product to buy.

Finally, AI technology used to automate tasks is likely the most mainstream understanding of AI, specifically when it comes to using machines to perform jobs at the same level of competency as a human.

In each category of AI adoption described, the technology used can be directly attributed to a business need — whether it’s dynamic product picking in a shipping warehouse or forecasting the demand of a product. This alignment is key to expanding company-wide AI capabilities and determining the data and infrastructure needed to support AI technology at scale.

From Raw Data To Training Data

There’s no doubt the e-Commerce industry generates massive amounts of data, from product SKUs to customer purchase history, but before that data can be ingested into a machine learning model, it needs to be put into a format the model can recognize. 

To ensure your organization is set up for success, here are a few things to consider when it comes to sourcing and labeling AI training datasets for e-Commerce.

  1. Data annotation can be time-consuming, but it’s important to balance labeling speed with efficiency and quality. Poor quality data labeling leads to flawed AI systems — establishing a quality rubric and quality assurance process helps reliably produce quality AI training data.
  2. The quality requirements for your data may vary depending on your model, but keep in mind that a diverse, high-quality dataset helps mitigate unwanted bias.
  3. Current, accurate and refreshed data is necessary to improve your model, however, sourcing data for a specialized use case can be challenging. Invest in a pilot project to determine the amount of data required to train and validate your model. Also consider partnering with a data annotation provider, so you can focus on model development without slowing down your labeling workflow.
  4. From data collection to model validation, there are a lot of moving parts in the training data lifecycle. Establish a feedback loop for real-time reporting of tasks, to ensure agility  in your AI pipeline.

Retail shopping, which once centered around shopping malls, department stores and big-box retailers, is moving toward the convenience and personalization offered by ecommerce shopping. The sheer volume of data produced by the ecommerce industry makes the possibilities for AI in e-Commerce almost limitless. Companies looking to adopt AI in their business processes should start with AI technologies that produce the greatest business value and assess cross-functional applications as new opportunities arise.

Wendy Gonzalez is the President and Chief Operating Officer and current interim CEO of Samasource, the trusted, high-quality training data and validation provider for 25% of the Fortune 50. In addition to holding two patents, she brings over 20 years of technology leadership and management consulting experience at EY and Capgemini to her leadership role. Gonzalez graduated from the University of Washington with a degree in Business Administration and Information Systems.

Feature Your Byline

Submit an Executive ViewPoints.

Featured Event

Join the retail community as we come together for three days of strategic sessions, meaningful off-site networking events and interactive learning experiences.


Access The Media Kit


Access Our Editorial Calendar

If you are downloading this on behalf of a client, please provide the company name and website information below: