In two weeks, a big global wireless carrier is partnering with an unnamed credit card company as part of a Black Friday campaign that merges its CRM data with transactional stats about what people buy. The pool of data will then be matched up with a marketing cloud’s database of 400 million people to find folks who are in the market for a new phone. It’s not until after terabytes of data have been analyzed that the wireless carrier’s campaign will earnestly kick in and the brand will begin pushing its ads to consumers, hoping that a super-specific group of people are more likely to buy phones than a mass group of people.
“We’re going to be marketing directly to those people starting at 7 a.m. [on Black Friday] and rolling throughout the weekend and into Cyber Monday directly because of the way this AI platform has ingested that data and made it work,” said David Steinberg, co-founder and CEO of Zeta Global, the marketing cloud firm behind the campaign. “The system will get smarter throughout the weekend—it will look at what people are doing as they open a message, the attribution module in the machine learning feeds that back into the master database and further fine-tunes the audience of people who will see that messaging.”
Steinberg declined to name either the wireless carrier or credit card company.
The scenario may seem wonky and hypothetical, but it’s an example of how marketers are increasingly pulling reams of data together and using artificial intelligence to better understand how people interact with a brand and how they shop as part of the so-called customer journey.
“An analyst can sit and look at a spreadsheet with 30 or 40 data points and then try to figure out how to use those data points to figure out what the next potential outcome is,” Steinberg said. “An artificial intelligence platform can look at tens of thousands.”
Steinberg claims that the AI and data-based processes generate up to 1600 percent higher rate of returns compared to generically targeted marketing. While traditional lookalike targeting creates broad nets for targeting—like women between 16 and 50 years old—AI analyzes and matches broader sets of data such as websites, weather and location for more specific targeting.
“When you have more specific targets and you can test faster, you can lower your cost of acquisition really effectively because you never have to deal with that big net that we talk about the first place,” said Chris Monberg, Zeta Global’s CTO.
Brands are also using data management and AI platforms to retain customers and analyze how likely folks are to stick with a brand.
Meal-kit delivery company Freshly has been working with marketing tech firm Optimove to pull together data and create targeted campaigns. Over the past year, the brand has gone from sending two targeted email campaigns per month to 40 customized campaigns per day.
“Acquisition and retention are very related to one another,” said Nate Champion, director of retention marketing at Freshly. “We certainly think about getting a lower customer acquisition cost and being efficient but what we really think about is how can we acquire the most valuable customer at the most efficient cost possible.”
Once Freshly acquires a customer, Champion said he’s able to track troves of data like how long someone stays on the site, rates their meal or changes an order.
Since beginning to work with Optimove, Freshly has increased its lifetime value of new customers by 19 percent, reduced its weekly churn rate by 22 percent and increased weekly order value by 64 percent.
The harder part is finding users and getting them to convert.
“You need to understand where customers are coming from, what they look like, how long it takes them to convert, what they converted on,” Champion said. “We’re not doing a great job on that component—I would say we focus more on simple stuff like the discount.”
Freshly is now testing a new tool from Optimove that combines both acquisition and retention, including serving CRM forms to people who visit the website but may not have converted.
“Well below 10 percent of the people who come through [the website] never convert but allowing smart segmentation at that level is very useful information for acquisition and conversion efforts because we understand which types of users tend to convert differently,” explained Pini Yakuel, CEO/founder of Optimove.
In addition to running an ecommerce site, beauty brand Glossier also has a blog called Into the Gloss that houses articles about beauty content. To understand how people are browsing between the two sites, Glossier worked with marketing tech firm Segment to pull the sites’ behavior and audience data together.
Glossier also collects data from social media, email and brick-and-mortar stores and needs one place to store all of its first-party stats, explained Peter Reinhardt, CEO of Segment.
“If you have five sources of data, you end up with 25 or 50 different connections that you need to create as an engineering team,” he said. “People wanting to use machine learning to build better products or better customer experience are discovering that they really need to focus on first-party data that they’ve pulled together from their own product and their own interactions with their customers.”
Don't miss Adweek NexTech, live this week, to explore privacy, data, attribution and the benchmarks that matter. Register for free and tune in.