- Real-time pricing information supplier Datasembly has rolled out an automated service that enables retailers and CPG manufacturers to match items they stock or make with products offered by competitors, according to CEO Ben Reich.
- The company’s new Product Matching tool uses machine learning techniques and natural language processing to rapidly compare publicly available details about packaged goods as well as fresh items like meat, seafood and produce, Reich said in an interview.
- Datasembly is launching the service at a time when grocers are stepping up investments in private label product lines and aiming to more deeply mine market data as they look to stand out with shoppers.
Datasembly has lately been receiving more frequent requests from retailers for ways to acquire more granular intelligence about the competition they face — a trend Reich attributed to the growing role private label products are playing as grocers face off with each other to attract and acquire customers.
“Something as simple as product matching — identifying when one product is the same as another product, or is a private label competitor of that product — is totally non-trivial at scale. And that's really where a lot of our customers are struggling when they want to understand the competitive landscape and when they want to track their competition,” Reich said.
The Product Matching system works by sifting through Datasembly’s growing collection of data about products retailers carry, including product titles, descriptions and category details. Datasembly uses those attributes and machine learning capabilities to generate hypotheses about what shoppers think makes products comparable with one another, helping retailers position and price items in ways that reflect people’s actual preferences.
"The rise of private label ... has just absolutely changed the marketplace dynamics in many categories, and it’s no longer the case that the private label alternative is just the budget brand that’s only bought by those who can’t afford the national brand,” Reich said. “This sort of sea change in how private label is perceived means that retailers need to track these dynamics with an increasing level of frequency and granularity.”
Capri Brixey, executive vice president of strategy consulting at Insite AI, which uses artificial intelligence to help CPG companies with category management, noted that data-mining technology can help retailers eliminate human error as they deal with questions around building their private label operations, such as which products to sell under their own brands.
“How much private label you put on the shelf or how much shelf space you dedicate to private label takes away from other brands. If you don't get that base elasticity decision correct, you’re leaving money on the table in an environment of narrow margins," Brixey, who is a former assistant category manager at Food Lion and most recently served as vice president of sales and customer leadership at Coca-Cola, said in an interview. Brixey joined Insite AI earlier this year.
Retailers who use Datasembly’s service can tailor their definitions of what makes certain products similar to each other, Reich said, adding that the company is striving to improve its ability to guide retailers in stacking up their products with offerings from competitors.
“There’s so many attributes that go into what constitutes products and what influences a consumer to think those products are competitive, including the brand, the size, the ingredients, ancillary attributes like organic and other descriptors,” said Reich. “Because we're collecting so much of this product information, and we're tracking it over time as well, the system is able to get smarter and smarter as time passes and hypotheses grow ever more accurate.”
Reich noted that Datasembly’s service can track and compare products on a store-by-store level, a capability that can help retailers fine-tune their product mix in direct response to local conditions. Datasembly does not provide the artificial intelligence technology that underlies the system, relying instead on existing algorithms to parse details on tens of thousands of products it has accumulated.