Sign up for our newsletter and get the latest big data news and analysis.

How Deep Learning Can Solve Retargeting and Reacquisition

In this special guest feature, Jeremy Fain, Co-founder and CEO of Cognitiv, discusses how with deep learning, marketers now have the ability to parse the browsing patterns of those who ended up purchasing items, and those who browsed but did not buy anything, and determine what the most likely paths to purchase are. Cognitiv is the technology powering IBM Watson Advertising products and is the first neural network technology that unearths patterns of consumer behavior so marketers can accurately pinpoint consumers who want to buy or further engage with their products. Cognitiv solves a formerly unsolvable problem for marketers: How to use Big Data to precisely buy ads that target the right consumers, at the right time, in the right place. It is the only solution that delivers automated, custom, self-learning algorithms to brands and the results have been seen to be up to 20X more efficient. Cognitiv was founded in 2015 and has offices in NY and Bellevue. Jeremy has over 20 years of interactive experience across agency, publisher, and ad tech management experience. Prior to founding Cognitiv, Jeremy led North American Accounts for Rubicon Project.

If you’re a marketer, conducting a campaign without implementing a process for retargeting is unthinkable. Why wouldn’t you retarget people who’ve already expressed interest in your product or brand? All you have to do is get a list of all of the people who’ve previously visited your website or clicked on one of your ads, and voilà! Let the conversions commence.

But, as marketers have discovered, retargeting is not as easy as it seems. Even if you can see whether someone has visited your website or not, and what (if anything) they put in their shopping carts, it doesn’t really tell you anything about the user itself, much less give you any insight into whether they will actually convert if you spend advertising dollars on them.

Essentially, marketers need to look beyond using retargeting in a blanket way, because realistically, the vast majority of people will navigate on and then leave. Even when someone abandons something in a cart, only a small percentage will ever come back and make purchases. The value lies in figuring out who are the people most likely to actually carry out the transaction, and then spending the retargeting budget on them. Although retargeting yields higher conversion rates than prospecting, the rates are still abysmal. Think about it like this: You’re taking a test, but you’re cheating, because you already know who’s pretty interested in your product. But despite all of that information, you’re still failing the test. That’s not something we should be patting ourselves on the back for.

With deep learning, marketers now have the ability to parse the browsing patterns of those who ended up purchasing items, and those who browsed but did not buy anything, and determine what the most likely paths to purchase are. Not only that, all of that information is particularly handy when determining the characteristics and behaviors of the people most likely to convert. In other words, using deep learning allows marketers the opportunity to get an even deeper understanding of what makes their consumer base tick, which comes in handy when looking to attract those without previous exposure to the brand.

As important as retargeting is, managing customer retention and reacquisition is equally important. It’s all well and good to get new customers, but if you can’t keep the ones you already have, it’s not going to do you that much good. For businesses with subscription-based plans, such as cellular service providers or magazine publishers, it’s a smart idea to keep an eye on customers who are thinking about quitting. Then you can offer them special packages (higher data allowances, special discounts, etc.) that give them another reason to stay around.

In the world of mobile gaming, this is an especially lucrative proposition, given the relatively high drop off rate that many games tend to have. If a developer had the ability to see if someone who had been playing the game fairly regularly in the past was experiencing fatigue or losing interest, they could offer incentives to keep players engaged. Or, let’s say they end up losing those players anyway. Now they want to bring those customers back, but some will never return. Which ones should they concentrate on getting back? This process is made much easier by the fact that they’ve already (clearly) exhibited interest in the game, and the data on their playing habits that’s already on hand. Deep learning can not only help the game publishers figure out the best way to target these users, it also gives them recommendations on how to best allocate budget, meaning that they no longer have to spend valuable resources on advertising to people with low likelihoods of converting, staying, or coming back.

Marketers like to be as efficient as possible, for obvious reasons. With limited budget to achieve your goals, why would you want to waste money by approaching someone who may have shown a modicum of interest, but will still never buy your product? The entire martech industry exists as a response to those problems and yet, it’s striking how many inefficient targeting technologies remain – simply because they are slightly better than doing nothing. The power of deep learning, on the other hand, is grounded in its proven ability to find customers by navigating through huge amounts of data to determine who exactly those people are. It essentially acts as a filtration system, separating out those who will convert from those who won’t.

Retargeting, retention and reacquisition are three fundamental issues that every brand has to address. Too many focus on such superficial criteria such as whether someone has previously spent time on their site without examining the behaviors that might lead someone to becoming a long-term subscriber or retaining their loyalty to a brand. This is an area ripe for disruption, and one that can clearly benefit from the advantages that using deep learning can bring. Apply deep learning, and suddenly, things will become a lot clearer.

 

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: