Online Retailer Delivers Email Success with Ople.AI

Case Study

Fortune 10 Subsidiary

Market Cap
$900+ Billion

Revenue
$200+ Billion

Company Description
As a top 30 eCommerce company, the retailer is a customer-centric and culture-driven company that specializes in eCommerce, Apparel, and Footwear.

“With over 50 different intelligent email streams, and a machine learning model needed to optimize and manage each email, it would take our entire data science team about two quarters to create every model we needed. With Ople, one junior data scientist finished all 50 models in about two weeks. Amazing!”

Key Achievements

  • One junior data scientist delivered individual models for 50 different customer emails in two weeks
  • The Ople-generated models were more accurate than would have been possible if developed by hand
  • With Ople, the company saved many months of data science man-hours with better results
  • Ople offers enough automation and ease-of-use that the marketers in the email group have taken over the model building for email campaigns
  • The success of the email project has created an environment where every team in the company are looking to use Ople
manu_bars

How to Optimize Email Communications

Over the years, the company had developed a number of intelligent automated email streams to send to customers and shoppers who interacted with the website. This included buyer reengagement emails that were sent if someone hadn’t shopped with the retailer for six months, or purchase incentive emails that were triggered if you leave an item in your shopping cart for more than 24 hours. 

 

Unfortunately, as the marketing team had developed a library of over 50 emails for different customer situations, they needed to put some structure and intelligence around how and when emails went to customers and site visitors. For example, they set a limit of six emails to be sent to any one customer in a calendar year. If left unmanaged, these emails could all be triggered in the first quarter, which meant you wouldn’t be able to communicate with the customer for the rest of the year. In addition, some customers don’t like certain types of emails, and receiving these emails might make them unsubscribe. These could include irrelevant offers, emails which are not personalized, feature the wrong types of products, or come from an unknown source (such as a list buy). Unsubscribes are bad for business as you can no longer communicate with a valuable customer. Not only does it decrease the lifetime value of the customer, but it makes it nearly impossible to win them back without investing additional budget at three to four times the cost.

retail_v_long

Marketing set out to optimize the efficiency of their marketing automation programs and with the help of artificial intelligence (AI), uncover an accelerated path to new customer purchases and maximizing the value of returning customers. The idea was to use machine learning to analyze the historical data for how customers and shoppers reacted to emails over time to find patterns for their behavior. These insights help the company put the right email in front of a customer when they are most likely to react positively (and purchase products).

With their requirements for an intelligent email strategy written, the email marketing team approached the company’s data science team to discuss the project. Based on what marketing was looking to accomplish, the data scientists quickly determined that each of the 50+ emails would require a separate predictive model. Using their current methodologies, and based on their current workload, the data science team informed the marketers the project would take months and that other parts of the business required their attention at the time.

Automated AI for Accelerated Success

Since the data science team was unable to assist at that time, the company’s email marketing team started to look for alternatives. One option was to hire data scientists, but in the current market it was a difficult proposition that would take months to deliver because of interviewing and on-boarding. And consultants could be brought on fairly soon, but they were extremely expensive and their manual methodologies would take months. 

 

After doing some research on automating the machine learning process, the team came across the Ople Platform. A true automated machine learning (AutoML) platform, Ople offered an end-to-end solution that helps to automate and accelerate the entire machine learning process to quickly complete AI projects. After reviewing the Ople Platform with the data science team, the company decided to pilot the email project with Ople. 

 

The company’s data science team offered one of their junior data scientists to work with the email marketing managers on the project. With the users identified, the Ople team provided the training needed to get the team up-and-running as quickly as possible. And the results amazed everyone!

manu_quote_purple

After reviewing the Ople Platform with the data science team, the company decided to pilot the email project with Ople. …with the users identified, the Ople team provided the training needed to get the team up-and-running as quickly as possible. And the results amazed everyone!

retail_middle

The Results

With the help of the Ople Platform, this motivated team delivered individual models for each of the more than 50 automated email streams. 

Each of the models predicts three different outcomes:

The Probability of Ignoring
If the intended recipient is likely to ignore a particular email, the marketing team can choose not to send it to them. With a hard limit of six emails in a calendar year, the company wants to send emails that have a high likelihood of being opened and acted upon.

The Probability of Unsubscribing
If the intended recipient is likely to ignore a particular email, the marketing team can choose not to send it to them. With a hard limit of six emails in a calendar year, the company wants to send emails that have a high likelihood of being opened and acted upon.

The Probability of Taking Action
In this best-case scenario, the company is able to identify those emails that are likely to spur the recipient to take the action that the company wants the reader to take. This could be purchasing a new product, taking advantage of a discount, or simply participating in the community the company has created.

Armed with these new models, the email marketing team was able to build an intelligent automated email program that addressed their customers and shoppers with the right message at the right time, increasing sales and engagement. And the company was better positioned to avoid sending emails that would cause a negative reaction, increasing customer satisfaction and reducing the overall marketing costs.

Building a Culture of Innovation

With Ople, the pilot team was able to produce over 50 models that drastically improved their ability to communicate with their customers effectively in just a few short weeks. And because of the simplicity of the Ople Platform, the marketing team is looking to use Ople on their own. The team has a great grasp of their business challenges and they know their data, so they should become proficient with Ople’s automation and end-to-end features very quickly. 

 

And with Ople, success breeds success. There are already efforts underway in other parts of the company to use the Ople platform to address other challenges. With the blessings of management and the data science team, the company is looking to embed AI throughout the organization to improve productivity, increase efficiencies, and drive revenue growth.

retail_bottom

The Ople.AI Platform

The easiest way to accelerate decision making and reduce risk with predictive analytics

With the Ople.AI Platform, you are only a few clicks away from building predictive models to derive optimal business recommendations with reduced risk.