In a presentation earlier today at eTail West in Palm Springs, Calif., J.P. Knab, chief marketing officer, and Cathy Bergstrom, group product manager, marketing content, both of Overstock.com, detailed how the home furnishings brand has built a technology stack that enables it to personalize its customers' experiences, particularly through the use of content.
Helping to create dream homes for all. That is the goal of Overstock.com, said Knab. But to make that goal a reality, the online retailer needs to truly be able to interact with consumers throughout their whole journey. And that means having a real-time dynamic content system.
"We needed to rip out basically our entire tech stack and replace it," noted Knab. He went on to explain the new system that Overstock has built, and the process that entailed.
Step one was getting to know our customers better, said Knab. That involved building a customer data platform (CDP) that leverages machine learning technology. In particular, Overstock focused on the following areas to get a better understanding of who its customers were and what they wanted:
- Recognize: Build a best-in-class ID graph.
- Listen: Capture, centralize and access data with proper permissions and opt-outs.
- Contextualize: Leverage real-time customer insights.
- Respond: Determine the best, most consistent omnichannel response.
"Machine learning is doing amazing things to help us understand customer behavior in real time, and help us come up with the proper response for each customer," Knab said.
With a customer data platform (CDP) built, the next step was ensuring that there wasn't a content bottleneck, said Knab. Overstock needed a content strategy that could do the following, enabled by technology:
- Be consistent: Have content that adheres to brand and voice guidelines across all channels.
- Be engaging: Use consumer insights to drive content creation.
- Be specific: Scale content creation and allow for rapid ideation.
- Be timely: Tag, store and score content in an asset data platform.
For Overstock, personalizing its customers' experiences meant creating a system that could deliver the right content at the right time to the right customer — leveraging data in milliseconds. According to Knab, that meant taking the following four steps:
- Identify journey. Engage with the customer when and where they are active.
- Determine method. Create or leverage ideal user experience for delivering the most relevant content.
- Respond. Deploy full customer journey activation.
- Learn. Create self-learning systems that feed back results.
Technology, Content Signs of Overstock Customer Love
Content is how online retailers talk to their customers, noted Bergstrom. She added that Overstock shows its love for its customers through the technology used to serve up content to them. And it's a lot of content. Bergstrom noted that Overstock has roughly 40 million content assets on any given day that could be shown to a customer based on the implicit and explicit data that the retailer has about them.
"We've built tools to show our customers our content and product," Bergstrom said. "A significant amount of machine learning allows us to do this. We need to convey the emotion and feeling of shopping for your home with the exact right piece of content — done in milliseconds and at a scale of millions. Talking to audiences or cohorts is no longer good enough."
Overstock has set the expectation internally that it has to be able to predict what the customer wants (through its use of implicit data), rather than just respond to what it's being told (explicit data).
A Case Study
Bergstrom ended her presentation by sharing an example of Overstock's real-time dynamic content system in use. Specifically, she cited how the brand is using machine learning to optimize email marketing campaigns.
When an Overstock subscriber now receives an email from the company, the content is loaded at the time of the open, and is based upon what the company knows about that customer — both implicit and explicit data. The data is not cached on a server and sent to an email service provider; rather, it's real-time data being leveraged at the moment of open.
This technology has driven improved performance for Overstock's email program, noted Bergstrom. The company has realized a 28 percent lift in email program revenue since launching its real-time dynamic content system.