Solving For 'When': How Data Science is Cracking Marketing's Toughest Problem

For marketers trying to figure out when customers are ready to buy, this startup has a few recommendations.

In marketing, as in showbiz, timing is everything. Even if you know what you want to say and who your audience is, when you deliver your message can be the difference between a sale and an abandoned shopping cart.

Data and programmatic tools can help marketers target and deliver content more precisely, but that is not enough without detailed information about where each customer is on their journey from consideration to purchase.

It’s a difficult problem to solve because of the variables and uncertainties involved. However, one promising approach is coming from an unexpected vector, deep in the murky waters of advanced data science.

The Element of Surprise

Element Data, a data analytics startup based in Seattle, is building tools to turn customer data into reliable, personalized recommendations. The platform evolved from the innovations of data scientist Charles Davis, a leading expert on predictive analytics for security and personalization, and has grown to incorporate social tools developed by former Microsoft senior engineers Vish Vadlamani and Phani Vaddadi.

Davis says Element Data’s technology framework, DICE, is designed to provide decision support and recommendations across a variety of enterprise and consumer scenarios, from product design, logistics and operations, to HR and finance. “Unlike search engines, which discover resources and information without regard to their relevance to the individual user or the context of the search, Element Data’s recommendation engine applies proprietary data models that present people with the best choices for them at the moment, particularly in high-stakes situations and complex decisions that involve a wide range of factors,” says Davis.

Though the platform was not designed specifically with marketing and sales applications in mind, Element Data has made several strategic acquisitions in the past year, including Vadlamani and Vaddadi’s startup PVCube, that enable it to determine when customers are moving from one stage of the journey to the next — and when they are showing intent to purchase — by identifying typical online and social media behavior patterns.

The Evolving Customer Journey

Customers go through different stages of consideration and exhibit different behaviors as they move toward a purchasing decision. When marketers can provide them with the right information at the right time, they can move that process forward toward a sale, and even recruit the customer as a “brand ambassador” and advocate by closing the loop with excellent service. But if they miscalculate the customer’s state of mind, they can send the wrong signals, appearing unhelpful, pushy, or irrelevant.

Customers go through this decision process on almost every transaction, but the intensity of consideration increases with high-stakes purchases (cars, appliances, major vacations) and the cycle compresses with high-frequency transactions (trips to the grocery store, routine online purchases).

Most sophisticated marketers understand the customer journey as it relates to their products and have content and channel strategies geared to the different phases. The challenge is that every customer journey is influenced by demographic, economic, social and psychological factors unique to each customer. Even with the ability to deliver personalized content at scale, it is very difficult to identify exactly where each customer is on their journey at any given moment.

Evolving Technology Solutions – Big Data and Machine Learning

Recent, rapid innovations in the field of machine learning are transforming how marketers plan and implement campaigns in both the B2B and B2C realms.

Machine learning can identify patterns in large bodies of customer data that correlate with tastes, interests, and buying behaviors, even if there is no intuitive causal connection between the customer attributes and the expected behavior. With machine learning algorithms in place, a marketing automation system could discern that a 24-year-old single female who likes Ryan Gosling, Adele, and the Wynonna Earp TV show, lives in a suburban zipcode, recently Instagrammed a picture of her new puppy, and just got promoted (according to her LinkedIn profile) is likely to be in the market for a new car, because that is the case in 80 percent of similar patterns.

Data-based targeting adds much-needed precision to campaigns, and benefits consumers who see offers that are more relevant to their needs rather than a lot of random offers and junk. This level of precision helps bring marketing closer to one-on-one sales, but it is still not quite good enough to replicate the intensity and personalization of a sales experience at scale.

The difference is timing.

A good sales representative can target a prospect from an initial encounter, taking note of their personal attributes, clothing, body language, and other indicators along with whatever might be in their CRM data — but can also continually adjust the sales pitch according to clues, signals, and responses. One or two questions rapidly reveal where the prospect is on the customer journey, allowing the sales rep to judge what kind of information or offer would best move the process along toward a close.

Online shoppers do not expect and do not always welcome personalized sales engagements when they are still in the initial stages of their customer journey, and may bounce from pages that interrupt their research and data-collection with an ill-timed pop-up or sales chatbot if they are not ready to buy. By the same token, merchants might lose a sales opportunity if they do not provide interactive Q&A, upsell, or cross-sell information or a promotional discount to a customer who comes to the site ready to buy if a single condition is met.

Building the Next Generation Customer Intelligence Tool for Marketers

Element Data believes their DICE decision engine provides the capabilities to tailor marketing campaigns to the specific needs of each customer, without creeping them out or driving them away.

“The utility of this solution depends on the software’s ability to accurately infer the user’s own priorities and values as a filter for determining appropriate choices,” says Vadlamani. In certain cases, users declare their preferences explicitly by answering a series of questions. The platform supplements user-provided information with additional data drawn from social media, search history, and other sources, providing a much deeper and richer view. “By comparing an individual’s pattern of choices with billions of others, the platform can offer guidance with greater confidence, accounting for factors that the user may not be aware of themselves,” he says.

The DICE platform’s ability to apply predictive analytics to individual behavior based on pattern matching and social graph data makes it an ideal solution for marketers looking for the missing piece of the customer journey: identifying the moment when customers show intent to purchase.

DICE Social, one of the platform’s service modules, is already optimized to support this use case. “DICE Social includes tools and dashboards that allow sales representatives to survey the social media landscape for active prospects, using the platform’s advanced data models to spot clues and patterns of customers moving from consideration to intent in the real-time social data stream,” says Vaddadi. “It can literally build lead sheets from nothing, harvesting bales of needles from the world’s biggest and most complicated haystack.”

Sales representatives can manually follow up on these leads with a personal approach, maximizing the odds of conversion with the right combination of information, promotional offers, and relationship value, ideally timed to close the sale. In B2B engagements, this can dramatically lower the cost of customer acquisition, discover best practices around conversion, reduce churn and transaction abandonment, and increase the average value of each deal.

“These capabilities promise significant benefits and rapid ROI for sales organizations, but DICE offers truly transformative value at scale,” says Davis.

The output of DICE Social can be exported via API to a programmatic marketing platform that can serve personalized content to millions of customers based on behavior patterns that indicate their precise location on the customer journey. Brands can not only present the right kind of content through the right channel at the right time of day to each prospect, they can respond to each customer’s individual priorities and stage of consideration, treating each marketing encounter as a personalized sales engagement.

Davis says Element Data is still building out the core capabilities of its platform and is working with thought leaders in the marketing industry to optimize its solution for real-world business needs. “We have a tool kit but we need input from experts to make sure we are focused on the right areas,” he says.

In other words, the makers of the recommendation engine are looking for recommendations!

Now Reading
Solving For 'When': How Data Science is Cracking Marketing's Toughest Problem