Using artificial intelligence, retailers can now pair website visitors with the perfect product for their purchase preferences—at the perfect time.
The result?
Increased revenue, more sales and better marketing ROI.
Those are benefits that AI platform Bluecore says it produces for retail operations, using robust AI and machine learning technologies. We spoke with Bluecore CEO Fayez Mohamood to learn more about the product.
Bluecore is a retail marketing technology company that works with over 400 retailers to transform customer communications into next-best actions by activating insights at the intersection of customer identity, onsite behavior, and, most uniquely, retailers’ product sets.
We use artificial intelligence to evolve retail marketing beyond tasks and toward strategies that drive high-value business outcomes (e.g. increased revenue, reduced margins, and optimal operational efficiency down to the product level).
We’ve built our foundation on a basic premise: customers respond positively to messages featuring products that are relevant to their specific interests and unique position within the buying cycle. Being able to pair a specific customer with their unique one-to-one interactions with products at this level of precision is not simple, but it’s our specialty.
Our technology first has to understand who each customer is and at what point he or she is at in their journey. It then needs to know exactly what products s/he has engaged with (e.g., viewed, clicked on, searched for, added to a cart) or purchased. And it has to do all of this at scale: across our combined client base, we manage more than 400 million unique customers and a product set larger than Walmart and second only to Amazon.
By tying a retailer’s behavioral and product data to specific customers, we’re able to paint a comprehensive picture of each customer and their historical relationship with a retailer’s products. Our technology uses this highly nuanced understanding of each individual customer to begin predicting his or her future relationship with specific products at a very high-level of accuracy.
Our platform’s AI layer can identify, for example, which customers have affinities toward certain products, even if they’ve never viewed those products. It can determine if a customer has a price sensitivity and therefore needs a discount to convert, or if that customer will happily pay full price (and we should therefore drive conversion without offering a discount). It understands who is most likely to convert and can direct advertising dollars towards this very specific audience rather than hitting a broad swath of people who may have a low likelihood of buying. The AI can even go as deep as pinpointing if a customer is at risk of churning, and if so, formulate the best method of reigniting that customer’s interest.
As with any legitimate machine learning model, ours gets smarter as it is fed more data, so the more a retailer can engage customers and drive them back to the site to interact with products, the more precise its predictions become. We credit our high level of replicable success to an elite and diverse group of data scientists who have deep experience training models and evolving them for better interpretability.
There is a lot of talk about general AI software being able to achieve any goal put in front of it, which reveals an underlying belief that AI can think and apply reasoning. This is not the case.
AI companies limit themselves when they forget this and try to tackle multiple disconnected problem sets simultaneously—or, act like traditional software companies and go broad or horizontal across industries. The key to good AI is focusing it on—and training it to master—one very specific problem set for a specific industry vertical, use case or user type.
This problem set should be characterized by multiple variables, each of which are changing in real time, at scale, and influencing the overall equation each time they do. It should also have clearly defined parameters. AIs need boundaries.
This is why it’s important to pick a vertical industry, a particular use case or user type. For us, that’s retail. We understand on a granular level what retailers are trying to achieve—increased lifetime customer value, higher revenue, increased profitability, reduced churn—the challenges that stand in their way and the numerous routes to achieving each of these goals. This has allowed us to tune our machine learning algorithms to optimize for exactly those outcomes and bypass those obstacles.
The data we deal with is very different than the type of data a financial company deals with. The customers are different, and how they buy is different. A machine learning model has to take all that into account. AI will continue to be limited if AI developers don’t recognize the need to specialize in early days.
There is a huge opportunity for AI to influence how brands connect directly to individual consumers at scale. It is not—nor never will be—humanly possible for retailers to intimately understand the preferences and future needs of each and every one of their customers and visitors. Yet, this is what consumers have come to expect. If a consumer engages with a brand even once—and then that brand targets them with an irrelevant message or product the next time they come into contact—the consumer questions their importance to the brand.
AI can also make business decisions that marketers can’t for lack of visibility into massive data sets. For instance: “This set of products is not moving fast enough, but given a limited budget and a specific email list of customers who meet X, Y, Z qualifications, we can meet sales goals while maintaining a high level of relevance with the customer.”
First, we’re the only system in the world that understands and can activate each product in a retailer’s product catalog, offer visibility into all customer interactions associated with each product, and connect all of that with individual email identities, at scale.
Second, we’re the only solution that’s then able to tie these customer-product interactions to real-time analytics, allowing brands to respond to customer and product-specific events on their sites and predict the best-next action based on the data.
Third, we focused heavily on making deployment easy for the end user (for real). A retail marketer’s first question is usually, “What’s a quick win I can get under my belt?” so it’s sometimes hard to sell “easy integration” as a primary value proposition. But as soon as a marketer is hooked on the business value, their second question will inevitably be, “Oh my god, is this going to be a total pain to stand up?”
We built the company on retailers being able to get enough data to start working right out of the gate. We have a team of what we call “forward deployed engineers” who know how to map the structure of a retailer’s website into something that is actually usable. There is a lot of really difficult stuff that we handle for you. We know how to scope it and we know how to launch in 45 to 60 days.
From a technical perspective, the ideal Bluecore customer is an online retailer with at least 50,000 to 100,000 monthly unique visitors and at least a couple of hundred products on the site. This ensures that we can pick up enough data to make the platform is valuable for the retailer.
From a more qualitative perspective, the marketers and a teams we work with are some of the most agile and curious professionals in the industry. These teams are willing and eager to test and iterate within the platform and they work closely with our customer success team to ensure they’re providing the best experience for their customers, driving the most revenue possible for their businesses, and doing so with the most intuitive and seamless workflows.
Some of the primary use-cases include: acquiring net-new customers (converting browsers to purchasers), converting one-time buyers to repeat buyers (driving second purchases), converting lost-buyers, and helping customers discover new products.
Those are the types of installations that make internal AI champions look good—the ones that allow people to spend more time on strategic thinking as opposed to doing tasks. Traditionally, the goal of enterprise software has been taking something that exists and making it more efficient. People are mistaken when they think AI-based enterprise-wide software completely changes everything. It’s not about new things you can do, it’s about doing things you already do more efficiently.