Machine Learning is opening up opportunities to think more predictively and strategically. Vidora is making Machine Learning a reality for anyone in any business. With Cortex, anyone can build an ML model. Optimize your business in just a few clicks via a simple and intuitive interface. A Vidora retail partner recently used Cortex to boost email revenue.

It is difficult to accurately target the right users with marketing campaigns specific to certain categories of products. Whether your categories contain eCommerce products, media content, or otherwise, marketing success hinges on the targeting users interested in a particular category, without spamming those who are uninterested. In this case study, we will walk through how Cortex allowed one of our partners to run category-specific targeted marketing campaigns powered by machine learning, and increased email revenue by over 400% in the process.

The Problem

Our partner is a large global retailer with a significant and growing digital presence. As customer purchase behavior increasingly migrates online, our partner is focused on maintaining a unique voice in a crowded eCommerce landscape.

Our partner’s marketing department often relies on promotional emails to stay top of mind and encourage transactions. Email marketing is a core component of their business. They consider email particularly important when promoting new products from a specific category to their customers.

One of the challenges that promoting specific categories of products poses is the risk of emailing too many users about products from categories that aren’t relevant to their interests. Doing so will not only harm cost of acquisition, it could also annoy customers with no interest in that product category and put them at risk of churning. The key challenge for our partner was finding the right users to target in order to maximize revenue without alienating uninterested users.

The Cortex Solution

Cortex offers a solution to this challenge. By placing machine learning technology directly in the hands of our partner’s marketing and business intelligence teams, Cortex allowed them to predict exactly which users are most likely to transact for a given category of items.

As a general rule, the more data available, the better Cortex will perform. But don’t worry about manually structuring or curating your data. Cortex will automatically determine the right data to use when building models. It does this by relying on various automated feature engineering techniques.

We collected data such as:

  • Integrated with real-time onsite behavioral data (streaming of online behavioral data directly to Vidora)
  • Daily ingestion of Purchase Data, App Data
  • Ingestion of App Signup Data
  • Daily Ingestion of catalog data

Our partner also was able to use Cortex’s easy integration tools to establish the infrastructure necessary to collect this user behavioral data on an ongoing basis in real time. Once Cortex was armed with enough data, our partner set out to run a targeted campaign for a specific category of products.

Cortex Pipelines

Cortex Pipelines offers a range of different model types to build depending on the use case. Our partner built an Event Time Series model to predict a transaction event within a specific window of time.

email revenue

Cortex Pipeline Predictions

Once you build your model, you face a tradeoff. On one hand, you need to reach enough customers to make your campaign a success. But you must also mediate the risk of bothering those uninterested in making a purchase. With Cortex Predictions, our partner was able to explicitly target the right number of users. They reached an estimated 5% conversion rate by selecting from a simple plot. They then downloaded the results from Cortex Predictions directly into their email marketing. Now they had the tools to optimize which users to target with promotions.

email revenue

With Cortex’s point-and-click interface, our partner was able to accomplish this whole process in minutes rather than months, and without the involvement of any data scientist intermediaries.

The Results

Our partner saw a 400% increase in average revenue per email sent using Cortex to target the right customers with its distinct offers. Cortex predicted which users were interested in making a purchase in a particular category during any given week. Our partner also significantly reduced their cost of acquisition. They also reduced risk of churn for users uninterested in that product category.

Summary

eCommerce companies can think about their product and marketing strategies in more specific, forward-thinking and strategic ways than ever before. This is thanks to Machine Learning. Cortex makes ML a reality for anyone, not just those with years of expertise in machine learning. Using Cortex’s predictive machine learning models, our partner increased average revenue from each email by 400%, and your business could see similar results.

Want to know more? Contact us at info@vidora.com!

Want to Learn More?


Schedule a demo and talk to a product specialist about how Vidora’s machine learning pipelines can speed up your ML deployment and ultimately save you money.