Artificial Intelligence to predict customer behaviour

 In Geen categorie

Artificial Intelligence to predict customer behaviour

Loyalty Lab Case Study, AWS Marketplace

About the Company

Loyalty Lab provides data-based marketing services, both to large enterprises and individual retailers. The company analyzes its clients’ customer data to find correlations, build profiles, identify targets, and design more effective marketing and loyalty programs. Headquartered in the Netherlands, the company has 55 employees based in three countries who serve over 2,000 customers in eight countries throughout Europe.

The Challenge

Explosive growth in the devices and channels that people use to interact with the companies Loyalty Lab counts as customers has led to correspondingly explosive growth in available data points. This became a challenge for Loyalty Lab data analysts.

“The amount of data we had to work with grew day by day—not the number of records, but the number of data categories available for each record,” says Dick Koers, a solution architect for Loyalty Lab. “The more data categories there are, the harder it is for people to find all the correlations.”

Loyalty Lab knew that machine learning technologies were well-suited to help it overcome this challenge, but the company lacked expertise in these technologies. It seemed there were two main options for obtaining the machine learning–supported analysis capabilities that would be vital to the company’s future. One option was to staff a new AI data science department to build and run an on-premises machine learning solution. The second was to license a managed solution for machine learning that could automate the steps that would otherwise have to be performed by data scientists.

Neither option was appealing. “Staffing a new AI data science department would have required us to replace more than half of our current data staff, which would have increased our personnel costs and been very disruptive,” says Koers. “And as a project-based company, we try hard to minimize costs that won’t always be billable to a specific campaign or initiative.”

Loyalty Lab wanted a machine learning solution that could be operated by employees without specialized AI data science expertise and that offered pay-as-you-go pricing.

“ in AWS Marketplace lets us concentrate on results rather than technology.”

Why Amazon Web Services

By early 2017, Loyalty Lab had not found a solution that met these requirements and was on the verge of staffing a new AI data science department, despite the drawbacks. Then the company learned that a machine learning solution offered by PredicSis, called, had become available in Amazon Web Services (AWS) Marketplace.

Loyalty Lab quickly saw that, procured through AWS Marketplace, would be an ideal solution for two main reasons.

First, for an AWS customer like Loyalty Lab, easily overcomes one of the biggest obstacles that machine learning solutions can encounter: the difficulty of obtaining live access to the necessary data. Loyalty Lab has relied on AWS services, including Amazon Redshift and Amazon Simple Storage Service (Amazon S3), since 2011, and in AWS Marketplace can connect to and start analyzing data stored this way in seconds.

Second, because it was available in AWS Marketplace, offered precisely the pay-as-you-go pricing model that Loyalty Lab wanted. “With that pricing model and the simplicity of, it was so easy to just try it out,” says Koers. “As soon as we did, we could tell that had the power we needed.”

To get started, Loyalty Lab specified which version it wanted to use, the AWS Region it needed, and the size of the Amazon Elastic Compute Cloud (Amazon EC2) instance that met its requirements—right from the page in AWS Marketplace. One click later, Loyalty Lab had launched as an Amazon Machine Image (AMI) and was ready to run its first PredicSis analytics initiative, a lead-qualification project for a major European automaker.

To identify which leads—in this case, people who had either requested brochures or taken test drives—were most likely to buy a car, a Loyalty Lab data analyst uploaded past lead-conversion data to Amazon Relational Database Service (Amazon RDS). Then the analyst used in AWS Marketplace to filter the data for features that correlated to conversions. After 40 minutes of manual work and a few seconds of application runtime, delivered a predictive model for analyzing the new leads dataset and could then be shut down. “For AI data scientists to design and run this same initiative would have taken at least 30 hours and cost many times what it cost us to use, and the results would have been nowhere near as detailed,” says Koers.

The Benefits

By using in AWS Marketplace, Loyalty Lab can deliver more value for its customers—simply, quickly, and at a very low cost. “Frankly, when we tell our customers what in AWS Marketplace can do for them, they can be skeptical,” says Koers. “Then, in just a few minutes, we can show them a beautiful output with results they never expected, and that skepticism turns to amazement.”

Loyalty Lab is able to accomplish this by using advanced machine-learning capabilities, despite employing no data scientists specializing in AI technology. “ is no more complicated to use than basic spreadsheet software,” says Koers. “Because it’s accessible to marketers and other business users, without requiring any IT involvement, in AWS Marketplace lets us concentrate on results rather than technology.”

Many companies would willingly pay a premium to amaze their customers this way, but the pay-as-you-go pricing of through AWS Marketplace meant that Loyalty Lab didn’t have to. “One of the strongest arguments for using through the AWS Marketplace is that your costs are tied to your actual usage and end up being so much less than licensing a managed solution for machine learning,” says Koers. “It’s really the icing on the cake in terms of how it augments the AWS services we were already using.”

Jean-Louis Fuccellaro, PredicSis chief executive officer and cofounder, says that Loyalty Lab’s experience is typical of its customers who obtain through AWS Marketplace, which is why this offering is now central to the company’s growth strategy. “When you look past all the hype about machine learning, the real question is how it can help you do real business,” he says. “The benefit of being on AWS and then adding is that the data is right there, you don’t need IT assistance, and any business user can try it to gain accuracy and speed. The democratization of machine learning has always been our central goal, and bringing to AWS Marketplace has turned out to be the perfect way to achieve that.”

Learn More

Learn more about AWS Marketplace Machine Learning and Artificial Intelligence.