We help retailers survive and grow in the rapidly evolving data-centric world. Today’s market forces include an accelerating shift to e-commerce, rapidly emerging fast fashion players, and the growing impact of high-scale players like Amazon, Alibaba, and Walmart. These data rich companies are amassing a growing advantage, given that data density is an essential ingredient to AI and sophisticated personalization. For example, Amazon is presently selling over 75 items per year to its Prime customers, who now number over 100 million.
We’d like to help all retailers achieve greater levels of digital success. We’re a behind-the-scenes ally of retailers, neutralizing the retail playing field by helping our customers overcome the data advantage possessed by their competitors.
At PreciseTarget we think a world with one dominant retailer would be a bad thing. We’ve built a cross-merchant solution that leverages the data and capabilities of hundreds of retailers. We’re a data science company focused on solving a retailer’s sparse customer data problem. In this paper we’ll tell you about an entirely new approach to digital marketing data.
The Goal: Build an AI System that Replicates Human Decision Making
Our R&D research began with an analysis of over 5 Billion consumer transactions. The transactions were provided by our retailer partners who share a passion for winning in the digital ecosystem. The provided data included a unique identifier for each consumer, the SKUs purchased, and the meta-data about each purchased item. We used sophisticated machine learning algorithms to ‘machine learn’ about these transactions. Think of it as a research project enabling us to virtually peer into the closets of over 60 million people.
Our data science research uncovered many surprises. One was that consumers have surprisingly weak loyalties to individual product brands. This doesn’t mean that they don’t like a brand they previously purchased; rather the data clearly indicated that they are equally open to other brands that match their taste. This is particularly true if the other brands are similar. This unexpected picture drove us to deeper research. What drives the decision making of a consumer if product brand isn’t the key decision driver?
A goal of our machine learning system is to learn the attributes of what people are buying. Our retail partners provide the meta data associated with the purchased products, which for a dress could include the color, size, style, fit, fabric type, and cleaning instructions. We saw that while people lack strong loyalties to a product brand, they consistently bought products that were similar to their previous purchases. Said another way, the dresses or jeans in a woman’s closet have clear similarities to others in her closet. We began mapping these similarities by examining the attribute similarities shared by the products. If she has a collection of similar jeans, what attributes make them similar? Our transaction files included 25 to 30 ‘features’ per product. Aside from color, size, and fit, the system was looking at details like the types of fabrics, country of manufacturer, and even the colors used in the product’s buttons.
This led to PreciseTarget creating many subsystems, including arcane things like product similarity matrixes, and taste scoring hierarchies. From a practical standpoint, these subsystems enabled us to create the retail industry’s first taste graph. The taste graph is a profile of each consumer’s taste in each retail category. As data scientists we’re compelled to prove whether alignment with a consumer’s taste changes outcomes for retailers. Our testing methodology was to compare a consumer’s likelihood to purchase an offer based on their taste, versus repurchasing a previous brand. The testing illustrated consumers were 50% to 200% more likely to buy a product matching his or her machine learned taste. This testing validated that a high-scale taste platform would make an important contribution to the state-of-the-art. [question – aren’t they buying a product that matched with our assessment of their taste? I think we should take a little more credit here}
Privacy mantra: Serve Customers, don’t exploit them.
We’re living in a world of consumer’s being exploited by marketers. It has been said that if the product is free you have become the product. We don’t think this is a fair deal. We’re admirers of Netflix and Spotify who use of aggregated data for the purpose of empowering consumers, rather than exploiting them. Society is nearing a breaking point on being exploited by their own data, and we believe tomorrow’s winners will be on the side of the consumers. If you told consumers Spotify and Netflix will no longer deliver personalized music and movies there would be great disappointment. People are fine with opt-in aggregated data being used for the good purpose of empowering them to make better decisions.
PreciseTarget has developed a Taste system that makes consumer exploitation impossible. Spotify and Netflix don’t sell data about what you’re listening to or viewing, and similarly, we don’t sell data about people and their purchases. Our taste system is designed such that no participating party, including PreciseTarget, can identify a consumer in our data. Further, we made it impossible to identify a retailer in our rich data set. Our Trusted Exchange System uses a secure third-party as the exchange agent with retailers, with only this trusted party holding the data match key. Our taste system uses synthetic identifiers that can only be translated by the trusted intermediary.
We’ve passed rigorous evaluations by the risk management lawyers at major retailers, major credit cards, and the largest advertising networks. Our data has been certified for use in these environments given that there is no connection to PII, nor an ability to decrypt an identifier.
Our Business: Helping Retailers Build Profitable Customer Relationships
We help retailers become more like Netflix and Spotify, delivering better end results to the consumers. Consumers are driven by their tastes, whether they’re selecting a movie, a music playlist, or purchasing an outfit on an ecommerce site. We’ve designed a system with only one purpose: deliver better consumer experiences based on the customer’s taste. We’re flattered when people refer to us as the Spotify of retail.
You can learn more about PreciseTarget here. Our CEO personally responds to all messages and you’re free to contact him at firstname.lastname@example.org. Feel free to send tips, ideas, or ways we can serve the retail community better.