Retailers are often severely affected by poor-quality data. Bad data can have a negative impact on business operations, customer satisfaction and sales.
Retailers’ bottom line can be negatively affected by bad product data. According to an information technology firm Gartner The average cost of poor data quality to organizations is US$12.99 million per annum. Long-term, it can have a compounding effect on revenues. Bad data can lead to poor decisions, as well as increasing the complexity of data eco-systems.
SaaS-based product search and discovery platform for e-commerce helps retailers understand the impact of inaccurate data. GroupBy In September, we hosted a webinar with Google Cloud Partner Sada Online-commerce firms Rethink Retail. The event, entitled “Bad Data Big Trouble: Turning the Corner on Poor Quality Product Data,” explored how businesses could use AI to enrich their data, improve product discovery and search relevancy, increase customer satisfaction, and reduce operational costs.
This level of success comes from identifying and improving areas in product data. In order to achieve best practices, it is important to establish a standard data model, conduct regular reviews, as well as implement AI-powered solutions that automate cleaning, normalizing, and optimising product data.
AI-powered data enrichment, therefore, can enhance brand recognition, improve operational efficiency and fuel growth. Arvin Natarajan is GroupBy’s director of products. He says that poor-quality data affects almost every retailer, and impacts every application which relies on data.
He said that “long-term, inadequate data affects your customer experience, and ultimately, the bottom line.”
The expert said that advanced generative AI model trained on GroupBy’s global taxonomy can identify data issues, and revolutionize the product data attribution management.
Cloud-based product discovery using AI
GroupBy’s ecommerce search and discovery platform powered by Google Cloud Vertex AI gives retailers and wholesalers exclusive access to Google Cloud’s next-generation engine. The platform is designed specifically for ecommerce and uses AI and Machine Learning to process over 1.8 trillion events. It also gathers 85 billion new events every day from Google’s entire product suite.
GroupBy can deliver digital experiences that are based on a deep understanding and knowledge of the user’s intent. Natarajan stated that their partnership with Google ensures customers will benefit from any AI innovations Google may develop in the future.
Incomplete, inaccurate and inconsistent product information can hamper search and discover, leading to reduced revenue and customer loyalty. Natarajan emphasized the importance of AI for data enrichment. He cited a 20% increase in sales in e-commerce after optimizing catalog software data to improve search and discovery. While AI plays a crucial role in data enrichment, businesses can also leverage tools like catalog software to ensure data accuracy and consistency in their product catalogs. These tools can automate many aspects of catalog production, reducing manual errors and streamlining workflows.
Fake data can expose revenue losses
Retailers may not be able to detect bad data if they are using technology incorrectly or if it is not used correctly. Vinny O’Brien, Rethink’s E-commerce Strategy, recalled an instance from his early days at eBay. He showed how faulty indexing led to a loss of revenue due to suddenly invisible products listings.
Working with a partner was necessary to discover that eBay had failed to normalize product data. For example, if someone was searching for a Nike sneaker, but there wasn’t a capital N when the data was uploaded, then the product would disappear after the initial phase of the search.
This failure was not confined to this single product. This was a recurring problem for all retailers using the platform.
“So, you just disappeared. You lost around 30% of your search volume. After we fixed the problem (which was not an easy task at a company that size), we recovered revenue at about a 20% to 25% rate for organizations. This is especially true for those who had large catalogs. It is an area that has a major impact,” he said.
The Challenges of Dealing with Bad Data Isolated
Joyce Mueller, Director of Retail Solutions at Sada, says that the problem of bad data is not a deliberate attempt to lower priority of product data, but rather an unintended consequence. The problem has existed for many years.
Missing, incorrect, or incomplete fields lead to inaccurate data. She suggested that the data specs may be incorrect or inconsistent across SKUs. Mueller said that without clean data pipelines, the data we have isn’t as complete as we might like.
“This has mostly been a problem with back-end systems. Now, product data that’s not complete, accurate, accurately described, or with a good character and style actually causes problems for online shoppers. “It makes your products less discoverable”, she warned.
Standardizing data: The elusive goal
A one-size fits all standard is a losing strategy. Previous efforts have not been successful.
O’Brien stated that in 2010, the biggest ecommerce retail platforms forced marketers to use a standard data collection for each product. This was done to make products more visible. It was good to adopt that strategy until a certain point.
“I believe managing the size of data is a challenge when large companies give this kind of mandate,” said he. “It has to be accepted and adhered to by all.”
He added that the scale of data governance and management is enormous. Various industries come into play, whether it is business-to-business or business-to-consumer. Considering other compliance complications, these verticals may include medical products or food-grade applications.
Different types of industries have their own nuances. O’Brien believes that managing this on a large scale is extremely difficult.
Bridging the Data Management Gap
Natarajan said that he notices a gap when interacting with retailers and distributors during conferences. It is ultimately a gap that retailers also have to manage.
“It is difficult to manage data of this kind at scale. I believe that’s why we don’t see a standardization of data in all industries and verticals.
Sada’s Mueller stated that she was unaware of any retail sub-vertical that handled it well. She believes that digital natives are better able to handle it because they’re new.
“When you think about traditional retailers, we are thinking of systems that have been in place for a long time and don’t necessarily communicate. It’s more difficult for an incumbent to fix problems like this and form and shape themselves in such a way as to adopt new technology. She noted that they have a larger legacy and more technical debt.
Some industries have better chances of managing data, because their products are simpler. Natarajan says that you’d have less product attribution for some of these categories compared to more complex products like engines, machines, and other things.
He said that the less complex products are easier to manage.
AI Solutions for Data Enrichment
The experts on the panel discussed how distributors and retailers could become more aware of their actions to overcome the bad data issue.
- Audit product data starting with the most important categories.
- Use AI-powered data enrichment, cleaning and cleansing solutions to improve the product data quality.
- Assess the impact on metrics, such as revenue and customer satisfaction.
- Create a data governance system to ensure that product data is accurate and consistent in the future.
- Test out AI-powered tools for data enrichment to evaluate the impact on product catalogs.
- Determine a champion in the organization to lead the initiative. This could be someone from the product marketing team.
- Modernize data pipes and consolidate data on products into a central, cloud-based platform to enable advanced analytics and automation.