Powering Personalization at RichRelevance

Posted by on March 28, 2016

In our last post from ThousandEyes Connect SF 2016, we’ll dig into how a SaaS provider uses performance data to improve customer relationships. Kevin Duffey, VP of IT Operations at RichRelevance, shares his experience delivering personalized services to hundreds of retailers around the world.

Figure-1
Figure 1: Kevin Duffey presenting at ThousandEyes Connect SF.

Making Shopping Personal and Fast

RichRelevance is a personalization platform that counts many global retailers among its client base. It’s a level of personalization much more sophisticated than the logic, “people who bought this also bought that.” Take for example the Nordstrom website, where RichRelevance helps the retailer provide a feature called “complete the look.” As Kevin jokes, “If you’re someone who doesn’t have a spouse to dress you up in the morning, you’ll see a shirt, belt and pants that match.” At L’Oreal, RichRelevance powers their recommendation wall and on-site demos. This personalization is infused into a retailer’s merchandising, search and content, and spans website, email, mobile, call center and store channels. Kevin explains, “It’s not quite like Tom Cruise in Minority Report, but we’re able to give you an in-store experience where kiosks are personalized to your tastes.”

The RichRelevance Technology Platform

So how does RichRelevance power a service that serves over 1B recommendations per day to hundreds of retailers in over 40 countries? Kevin states, “I’m proud to say that our front-end services have been up 100% for 65 months. While our back-end does have maintenance periods, we’ve never missed a recommendation, even during Hurricane Sandy,” eliciting gasps from the crowd.

RichRelevance is a collection of applications built on a common platform, enabling customers to bolt on functionality, similar to the Salesforce Force.com platform. This platform is served from 13 data centers in North America, Europe and Asia that replicate information among each other. This footprint is necessary, Kevin explains, because he expects to serve recommendations “within 250-300 miles of a customer’s data center, translating to a latency of 60-120ms. Our content will typically show up on a page faster than the retailer’s own product. Retailers like that their customers can see something right away.”

Kevin’s team chooses data center locations by considering a combination of customer data center locations and network density and performance. “We use products like ThousandEyes to help make decisions about networks and co-location providers in the geographies we care about,” Kevin explains. “This has created a tremendous partnership with my co-location vendors. Rather than just taking measurements from a looking glass, I point a test into the environment and start seeing the real performance.”

Using Operational Data to Build Trust

RichRelevance is responsible for powering personalization on major retail sites like Barney’s, Neiman Marcus and and LL Bean. Outages and slow performance are not an option; these retailers would be missing out on revenue. So building a trusting relationship with these retailers is one of Kevin’s key tasks; customers have to be able to rely on personalization being delivered to their end users.

Kevin brings up an example from Thanksgiving 2015, “a kind of busy day for retailers. I had clients calling out of Europe saying that they weren’t seeing the RichRelevance recommendations.” Kevin knew that his data centers were operating fine. So Kevin opened up ThousandEyes, which he was trialling at the time, and saw a “big red dot in the middle” between various European vantage points and his Frankfurt data center.

Figure-2
Figure 2: Outage in the NTT network, affecting traffic to the RichRelevance Frankfurt data center.

“When you click on the red interface, you see that it’s experiencing 73% packet loss. NTT lost their whole 10Gb backbone network outside of Frankfurt. By looking at this,” Kevin recalls, “we were able to have our co-location vendors and customers reroute traffic around NTT within 30 minutes.” Most importantly, when he got a call from the CIO of a large European retailer, Kevin “was able to turn the situation around and build trust with my merchant, my co-location providers and my carriers. ThousandEyes paid for itself in one single event on Thanksgiving.”

Winning Customer Loyalty with Data

In another example, while working on a deployment for Tesco, the largest grocery retailer in the U.K., the load testing team was seeing some problems. Kevin’s team had deployed a ThousandEyes Enterprise Agent within the Tesco environment to help with pre-deployment testing.

As Kevin displays the data: “Yeah, there’s one little problem we’re seeing. There’s 100% loss to one of Tesco’s proxy servers.” After viewing the interactive links that Kevin had shared, Tesco’s team took this information and quickly discovered a database misconfiguration. Kevin had just found out that morning that the Tesco account had launched and was live. “Very quickly we’re able to sit down with a customer and do something that we weren’t able to do before. We’re able to walk customers through information such as DOM load time and page load time.”

Figure-3
Figure 3: Tesco’s proxy server issue during pre-deployment testing.

According to Kevin, “ThousandEyes has changed our world in the last three months. We can drill down into individual cases with our customers. Each merchant is different, and I can share a link and have a conversation with them.”

Interested in hearing more about how ThousandEyes is used in the wild? Check out the other talks from ThousandEyes Connect SF by Zendesk and Cisco.

Processing...