3 Detours to Avoid in Retail's Race to Win With Big Data

There's no question that the era of big data is benefiting retailers. With data coming in from dozens of points on a consumer's path to purchase, retailers are rich with information that can turn shoppers into buyers. The catch-22 is having access to so much data is often too much to handle. Trying to make sense of it all is overwhelming, and for most retailers, a seemingly impossible feat.

One of the major roadblocks inhibiting retail marketers from winning with data is the misconception that more complex data sets deliver more value. Many are easily thrown off course, spending valuable time, energy and resources to extract insights from obscure, complex or unproven queries, while much more telling information lays right within reach.

By avoiding these common data detours, retailers can focus their attention on actionable data that brings customers across the finish line:

Detour 1: Embarking on the search for "true" meaning. Most consumers start their online purchase path with a search; most retailers recognize that the information in that search bar can help them target the shopper with better ads and more personalized recommendations. All true. The problem is, some retailers are trying to extract meaning from search using natural language processing and are spending valuable time and resources looking for human understanding instead of valuable correlation.

When it comes down to it, words are data, and we should think of them that way. By adopting tools that gather search terms and make correlations between the words a shopper writes and what they ultimately buy, retailers can personalize the shopping experience without having to really understand what the searcher "meant." For example, data-crunching machines don't understand the meaning behind "red stainless steel gas BBQ." But because they track word data, clicks and purchases, the machine knows which products or content are related to the terms and can populate the shopper's screen with a grill that has the appropriate specs without understanding the words at all.

Detour 2: Hoping (purchase) history repeats itself. At first blush, it seems reasonable that if a customer purchased an item at your store, it makes sense to "personalize" their shopping experience by recommending similar or related items when they come back. However, beware of making assumptions about a person's behavior based on history alone. Case in point, this shopper who bought a digital jewelry scale on Amazon.com and wound up seeing product recommendations for a presumed "business" he wasn't in.

As humans, we're wired to look for causation — i.e., that one thing follows another. Unfortunately, just because a customer bought a scale two weeks ago, doesn't mean they're buying related items today. While history is useful, we can't forget about the customer's current intent — what they're looking for today. Shoppers will often tell you what they're looking for through search or as they start to navigate. These simple behavioral signals you already have can give you the information you need to deliver a better experience.

As it turns out, historical data is useful, but not perhaps in the ways you might think. History can tell you about a customer's brand preferences, spending power and other useful information. These insights can then be used to drive an even more relevant experience when you understand the intent of today's shopping visit.

Detour 3: Getting too social with sentiment. It's no secret that purchases stemming from social activity are skyrocketing, and everyone wants a piece of the pie. But the new trend of trying to pinpoint customers and make recommendations using complicated sentiment analysis and natural language processing isn't going to get marketers the results they want — at least not now. We're still at the very early stages of automating a way to extract feelings and meanings from social posts, and many of the methods are flawed or unproven.

Much like search, simple correlations and patterns in a user's network will get you a lot further — and with a lot less work — than trying to understand a shopper's true feelings about your brand or products. If you want to know what a customer is going to buy next or if they're about to unsubscribe, look at what their "friends" are actually doing, what products they own, etc. Simple aggregation of this available data can tell you which products or brands are most popular with a peer group. This data can be very helpful in determining the likely next actions of related users.

When taking stock of the massive data coming through on the path to purchase, retailers will do best by focusing on search correlation over understanding, purchase intent over history and social network behavior over semantics. By avoiding these detours and keeping your big data strategy simple yet actionable, you can get on the right track to data success. Check out the infogrpahic to learn more.

Dan Darnell is the vice president of marketing and product at Baynote, a provider of personalization software for e-commerce brands. Dan can be reached at ddarnell@baynote.com.