Welcome to the Oracle CX blog:
The latest in customer experience strategy, technology, and innovation.

Unlocking Greater Personalization with Modern Commerce Technology

Madhukar Kumar
VP CX Cloud

What is ecommerce personalization and what does it really mean?

Forget ecommerce for a second, and take a journey back in time with me.

Walking into a 1920s general store, chances are the owner knows me by name and knows what I bought the last time I visited. Maybe our families are close. The shop owner makes recommendations based on past purchases and a keen understanding of my personality and needs. You could argue this was a preferred shopping experience for consumers for much of the twentieth century.

Fast forward to the twenty-first century and we noticed that while more convenient, ecommerce by and large ushered consumers away from such personalized experiences. Then in the past few years we saw that consumers were beginning to want both convenience and personalized shopping experiences. And today, we are realizing this is the expectation.

Today both merchants and technology companies are working feverishly on a variety of personalization applications.

Advances in machine learning and data science in particular are helping merchants deliver the same level of personalization online a consumer may expect offline with a one-to-one engagement. Individualization of online experiences is now a reality and the consumer never has to leave their living room.

The age of data-driven, AI-empowered commerce individualization has arrived and is here to stay. Merchants can now leverage 1st, 2nd, and 3rd party data to increase repeat purchases, improve loyalty, and satisfy customers.

In the following paragraphs, I’ll uncover how modern commerce applications like Oracle Commerce Cloud help merchants achieve this level of personalization through artificial intelligence, machine learning, and data science.

How AI and data science help merchants achieve personalization 

Making a personalized recommendation is a challenge if you don’t know who your customer is. The first step is to have ample access to the right data.

Here are some of the data points currently used by a majority of digital commerce sites to personalize product recommendations:

  1. Location data - By looking at an IP address, the general sense of customer location is gathered and this data is often used for product recommendations by location.
  2. Purchase history data - This is one of the key data points used by a majority of ecommerce technology providers to recommend next purchase.
  3. Demographic data - This information is used to make product recommendations based on age, gender, affluence. Once the demographic, psychographic and behavioral data is collected data science teams look for correlations across location, purchase history, and demographics and use it to either make next purchase recommendations or find segments of prospects most likely to buy a product.

Having this data is an important start. However, there’s a very big challenge with this process as it requires manual methods to create rule-based personalization.

This makes complex personalization nearly impossible, as customizing rules and attributes can get very unwieldy, very fast. In the past several years, many data science and analytics teams have spent countless hours on developing algorithms and rules that will recommend products their customers are most likely to buy. Here’s how this generally works:

  • You define the attributes that indicate interest - collect the data then clean, process and ship it over to the analytics team.
  • The analytics team builds models including, “Based on this, people who buy X product, are also likely be buy Y product, and are in Z industry” and run their analysis. This analysis then drives the rules for personalization.

As amazing as these insights might be, they are still rooted in complex rule-based personalization. What if the market changes drastically overnight? Or a new manufacturer enters the market? You need to completely rebuild your rules.

Here’s how rule-less personalization with AI works

With modern commerce technology like Oracle Commerce Cloud, merchants can crunch 1st, 2nd, and 3rd party data, even among first-time site visitors taking advantage of Oracle’s Data as a Service.

By combining data sources in real-time artificial intelligence can run statistical modeling using machine learning libraries to uncover insights that would take a team of data scientists weeks to crunch. This rule-less approach works in real-time and adapts to every single customer. Furthermore, you can triangulate mobile devices and graph IDs, isolating customers across devices in their precise moment of need.

This moves merchants from rule-based systems that must be manually updated, to real-time recommendation engines rooted in artificial intelligence and machine learning libraries. And because Oracle’s AI strategy is application based, merchants can embed rule-less personalization in their site experiences with minimal effort.

Two advantages of rule-less personalization

The spray-and-pray marketing model has unfortunately been acceptable for decades. Single digit email open rates are still the gold standard. Even pay-per-click ads with a 2% conversion are considered stellar.

Ever wondered why so few digital commerce brands challenge this trend?

It’s because digital commerce technology is like online dating technology. You’re trying to match attributes and deliver a recommendation. The sheer number of attributes that must line up can be daunting. But whereas online dating sites can rely on their users directly sharing these attributes, digital commerce brands must rely on artificial intelligence and access to the right data in order to deliver the same experience. This data compounds and becomes more complex every day and is subject to ever-changing consumer shopping behaviors.

With rule-less personalization, matching up commerce experiences and moment-by-moment consumer needs is truly possible. Here are two major advantages:

1. You know you’re selling to the right audience:

Let’s say you’re selling a bike. AI and data science allow you to leverage a huge sample size to evaluate other customers that have likewise bought bikes. By running supervised machine learning models, you can determine in minutes which attribute combinations triggered a purchase. Then, all you need to do is go find customers with the exact same attribute combinations and you have a high-propensity-to-buy customer segment you could market to. Just like that, you know you’re selling to the right audience.

2. You convert more customers coming to your site:

AI helps you find the right audience, find the right match, and then personalize and individualize the experiences at scale. By starting with the most accurate snapshot of customer needs, your A/B testing and personalization efforts can further refine the experience thereby convert more customers.

Greater personalization through integrations

Back or front office, today’s ecommerce platform does not live in isolation. This means integrations are an essential component for any modern commerce platform.

Many small businesses get started thinking they only needed a payment method and a shopping cart. But when you break it down, there’s much more complexity in even the simplest platform deployments.

There’s the CMS, ERP, inventory management, product SKUs, security, and more. How do you develop continuity across systems and provide a frictionless customer experience when a vast majority of touch points are outside the ecommerce platform?  You need an integration strategy. What about email marketing? Integration. What about social listening, engagement, and tracking? Again, that’s a separate integration.

Merchants are struggling today because as they grow, they simply can’t scale platform integrations fast enough. Many are investing in Integration Platforms as a result. But, this can cost millions of dollars.

A better option is to invest in a platform that boasts leading integration capabilities and helps you build your customer experience platform modularly.

Oracle Commerce Cloud is part of the Oracle CX suite that offers an Integration Cloud Service (ICS) along with hundreds of prefabricated connectors to other applications and platforms, several of which are already hooked up to all of our CX products. This means you get up and running faster and in a more affordable fashion. Oracle Commerce Cloud offers prebuilt connections to Responsys, social, sales cloud, CPQ, and more.

The possibilities are truly endless with today’s modern commerce technology. Merchants can go from simple recommendation-based personalization to rule-less individualization at light speed. But they must choose a platform wisely. A platform that allows them to start small but is also scalable. A platform that has pre-built machine learning based recommendations with one of the largest third party data sources. One that comes with its own integration platform so that you can provide a seamless and friction-less customer experience across all touch points even outside ecommerce.

Looking to learn more about what’s next in commerce? Download The Executive's Handbook to Modern Digital Commerce >>


Join us for additional insights and best practices to develop personalized shopping experiences at Modern Customer Experience. It's the industry-leading CX event in Las Vegas:

Join us at the Modern Customer Experience event

Join the discussion

Comments ( 1 )
  • Deepak Pal Wednesday, October 24, 2018
    AI and ML help e-commerce businesses in doing personalization. It works on training models and uses historical data to analyse user behavior and predict it in real-time. Businesses can use those actionable insights to personalize the store for specific segments of user. I have written an article on uses of real-time analytics in e-commerce. Hope it will help your audience: https://www.iqlect.com/blog/6-top-ways-to-use-real-time-e-commerce-analytics-for-shopify-stores/
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.