Predictive Analytics Solve Retail’s Trickiest Problem: Customer Intent

Learn Aster
Teradata Employee

This article is originally published in insideBIGDATA.com

 

The notion of predictive data analytics as a game changer for retail is nothing new. We’ve heard this now for the past few years, with industry experts noting customer identification and personalization as the most obvious benefits.

 

It’s certainly a compelling prospect.

 

But when it comes to achieving that single view of customers – which is what would effectively drive that useful personalization – the rubber hasn’t really met the road. At least, not for most retailers that aren’t Amazon, not yet. As retailers look to implement the omnichannel experience, they’re still very much in the experimental phase, working to integrate online and offline information with the goal of really understanding their customers’ journey.

 

What we most urgently need, however, is a window into customer intent. What does the customer want? When are they likely to buy? What gets them into a store? Retailers have many questions but few answers. Offline data, specifically location data, gets us closer to resolving the most pressing of these conundrums primarily because it provides major contextual signals. Knowing, for instance, that a customer has visited a store is a significant signal indicating intent. If we can use this data, along with its online counterpart, we can accelerate our ability to deliver the right message at just the right time to the right person.

 

That’s what predictive analytics is really about: helping retailers forecast intent. If we can make smart, data-backed predictions on what customers are likely to do, and when, retail marketers can then deliver a true curated shopping experience.

 

Some predictive models are trained using visit data from trillions of events across millions of visits from thousands of locations. This kind of model makes predictions based on key data points about visitors including number of visits, days since the last visit, visit duration, number of locations visited, and more. Historical visit and behavioral data then helps the model refine its accuracy and deliver better predictions of future visits. This class of methodology delivers an actionable, analytical, and novel approach for retail marketers to identify and classify visitors, including understanding how often different groups of customers return.

 

Online marketers have used predictive analytics to predict where and why users click, but clicks and anonymous visitors don’t translate well into solid predictions on precisely who will visit a store and when. Predictive analytics that rely on location data, as well as online and other offline data, can really move the needle on this. If we know who customers are, and we leverage a large data footprint, we can help retailers specifically predict which shoppers will be visiting their commercial locations within a specific time period.

 

These data-backed insights can then be taken one step further. Marketers can use them to tailor and personalize their outreach to specific customer groups and even individual customers. It’s a much more effective use of marketing spend.

 

Imagine, for instance, you’re a marketer at a large national retailer. Predictive metrics will tell you which customers are least likely to visit in the next 30 days. You then target them with a higher-than-average coupon that expires in 14 days, a generous nudge that gets them in the door. Or imagine you’re a fast food chain – and predictive analysis has revealed a set of customers with a medium likelihood of visiting in the next 30 days using their loyalty program. The analysis inspires you to give a free coffee that month when these customers come for breakfast instead of lunch, and it works.

 

Predictive analysis enables customization and much better targeting. Indiscriminate coupon blasts waste effort and money with no upside in customer acquisition or retention. But predictive analysis is more akin to a surgical strike, enabling retailers to provide differentiated engagement campaigns that can measurably increase loyalty and attract new customers.

 

The retail industry needs to develop tools, similar to those that e-commerce deploys, in order to thrive. Although location data is – in many ways – nascent, it’s a powerful enabler of predictive analytics in the physical world. And as online and offline signals improve, and machine learning improves with them, retailers will be able to deliver the highly tailored experiences that save customers time, money and effort, while improving their brand reputation and revenue in the process.

 

 

 

---------------------------------

Rachel-Wolan.jpgIn this special guest feature, Rachel Wolan, Vice President of Product at Euclid discusses  what predictive analytics is really about: helping retailers forecast intent. If we can make smart, data-backed predictions on what customers are likely to do, and when, retail marketers can then deliver a true curated shopping experience. Rachel is a seasoned product executive, bringing over 15 years of experience in B2B SaaS product, engineering and analytics. She is responsible for the Product and Design team at Euclid. She received a BS from Northwestern University and an MBA from the University of California at Berkeley.

 

 

 

 

 

 

To read the original article, visit Predictive Analytics Solve Retail's Trickiest Problem: Customer Intent