Wayfair: Creating engaging, digital-first experiences at scale with Google Cloud

A new sale demonstrates the need for scale

Every retail organization understands the power of promotions. That’s why we have institutions like Black Friday and Cyber Monday. Wayfair, an online furniture and home goods retailer headquartered in Boston, Massachusetts, took things one step further. In 2018, the company launched Way Day, an annual sale event featuring deep discounts on many of Wayfair’s most popular product categories. The inaugural Way Day was an extraordinary success, but Wayfair knew it needed to make its infrastructure even more resilient to keep up with the company’s exponential growth. Wayfair carries a large number of products in order to serve its diverse audience. With more than 22 million different products from over 16,000 suppliers, the company needed a technology partner to help it expand beyond its on-premises data centers and scale for large shopping events.

Scalability is a big issue with on-premises data centers. To accommodate bursts in demand like those seen by retailers during discount events, you have to predict traffic well in advance, provision equipment, and ensure that the servers are available and functional when needed. If your predictions fall short, or if there are unexpected delays in the provisioning process, you may be unable to accommodate the increased demand. Even if you do have enough capacity for the spike, you end up with expensive equipment sitting idle during non-peak periods.

Wayfair needed responsive, scalable burst capacity for peak events like Way Day, Black Friday, and Cyber Monday. The company chose to embrace a hybrid cloud.

Tailor-made integration of public and private cloud

Wayfair is proud of its heritage as a tech company, with more than 3,000 in-house engineers and data scientists working on the development and maintenance of its platform. The engineering team wanted to select a cloud provider that would help enable them to build a solution that was a perfect fit for the organization. Wayfair chose Google Cloud.

“We really needed to have access to the raw infrastructure,” explains Steve Crusenberry, Vice President of Infrastructure and Platform Engineering at Wayfair. “An ‘as-a-service’ solution wouldn’t work. The ability to build to our own specifications was a deciding factor in choosing Google Cloud.” The Google Cloud team worked closely with Wayfair engineers to ensure that there was complete transparency as they worked collaboratively to create a hybrid solution. Using a variety of Google products, the teams created a system that facilitated on-demand scaling from their data centers directly to the cloud, while handling multi-petabyte data accessibility without compromising security.

“When we first started the partnership with Google Cloud, we owned and operated three data centers,” says Crusenberry. “Initially, we extended those data centers into Google Cloud to create a hybrid model data center. In the hybrid data center, we split the traffic for our website and all our logistics and advertising capability, and we used Google Cloud to burst when we had large sale days or when we did maintenance in a data center. Google Cloud functioned as a source of excess capacity for us to handle large load spikes.”

The first real test of Wayfair’s hybrid cloud came on Way Day 2019. Not only was the newly implemented hybrid cloud able to handle the increased load, it also did so without incurring excessive costs. In early 2020, Wayfair then decided to move more of their capabilities to Google Cloud, including the entire Wayfair.com storefront. Crusenberry says, “It worked, and it worked incredibly well, not only in terms of the overall performance of the site and all the technology, but also in the cost of that technology. The expenses were right in line with where we expected them to be and on par with running costs we’d accrue if we had used our own data center.”

Improved efficiency with ecommerce migration and analytics

Upon reviewing the data from Way Day 2019, leadership identified an additional opportunity for improvement. “We realized that we would actually be better off to split the infrastructures,” says Crusenberry. “We decided to push the Wayfair ecommerce engine (including many Windows Server-based workloads) onto Google Cloud. We could then use the extra capacity created in the data center for backend processing instead.”

Shifting the ecommerce engine onto Google Cloud created enough excess capacity in the data centers that Wayfair was able to reduce infrastructure spending. “Not only did we save money on day-to-day costs, we also saved money because we no longer needed to buy additional capacity for the applications that were still running in the data centers,” says Crusenberry.

With the updated hybrid cloud model, Wayfair was able to seamlessly accommodate large spikes in traffic, as evidenced by the success of Way Day and Black Friday/Cyber Monday events in 2019. This success continued into 2020 and beyond. Seizing another opportunity, the Wayfair team was among the first to use Google Cloud’s C2 processors. The increased processing power speeded Wayfair’s applications and also reduced its cloud expenditure because the company could manage the same throughput using fewer processors.

Wayfair also improved the quality of experience for both its customers and its suppliers by bolstering analytics. After the migration, Wayfair began using Google Cloud’s smart analytics platform to gain real-time insights into their data. Wayfair ingests data from Pub/Sub, which is then stored centrally, with unstructured data in Dataproc and structured data in BigQuery. Wayfair’s team takes advantage of Looker's powerful data visualization capabilities to create curated dashboards that provide insight at whatever granularity a given team member might need. This ability to make sense of the data has real value; customers benefit from Wayfair’s ability to use this data to make smart, personalized recommendations. Suppliers benefit because Wayfair can provide feedback and forecasting, helping improve inventory and better predict customer demand.

Wayfair, Google, and COVID-19 As planning ramped up for peak events in 2020, the world was unexpectedly brought to a halt by the COVID-19 pandemic. With technology at the core of its DNA, Wayfair’s workforce has long embraced many of the virtual tools that became essential when working from home. Wayfair was already a longtime Google Drive user, and in 2019 the company had shifted to Google Workspace for all its collaboration tools. Like many Google Workspace users, Wayfair encouraged employees to connect using Google Meet and Google Chat.

Because Wayfair is a powerhouse in the home goods industry, the company became an important resource during the pandemic. Wayfair has provided necessary supplies like bed linens to medical facilities, and in April Wayfair hosted a shopping event that raised $2.3M for the Feeding America COVID-19 Response Fund, going directly to supporting more than 200 food banks, 60,000 meal programs, and millions of Americans during the pandemic. Wayfair also has been able to support individuals who unexpectedly found themselves working from home.

The widespread transition to remote work meant that many people needed to create a dedicated space where they could work at home. Wayfair had warehouses stocked with office furniture, which meant that it was able to rapidly respond to this increase in demand. Crusenberry shares, “Our volume rapidly increased in just a few weeks—people turned to Wayfair as they explored how to use their homes in new ways for comfort and entertainment during highly uncertain times. We’re working relentlessly to continue to serve our customers across two continents as they navigate their new realities.” The uptick in sales was substantial—Wayfair reported 82.5 percent growth in U.S. total net revenue in the second quarter of FY20.

The hybrid cloud also was able to meet the increase in demand—so seamlessly that, at first, the Wayfair team was able to maintain a business-as-usual approach to software releases. Crusenberry says, “We used to freeze changes and stop software deployment if we reached a certain traffic threshold because we didn't want to destabilize the website while under peak traffic. In Google Cloud, we were able to reduce and eventually eliminate the need for change freezes leading up to big events. Our automation systems signal the cloud to scale on demand and therefore handle the peak load while remaining stable and performant. We no longer have the same concerns about a software release or infrastructure change adversely impacting our customer experiences.”

Collaboration and flexibility as differentiators

The collaborative relationship between Wayfair and the Google Cloud team was key to Wayfair’s success in moving to a hybrid cloud. “Early on, we were encouraged to have members of the Google Cloud team be embedded directly in our team all the time. We also pull in individual experts from Google Cloud when we adopt newer technologies or more advanced capabilities,” says Crusenberry. “They work like they're part of our organization. And on big sale days, we have members of the Google Cloud team in our command center almost around the clock. That collaboration gives us a safety net in case something goes wrong. It’s really enabled us to shift traffic to the cloud with confidence.”

The close collaboration between Google Cloud and Wayfair meant that the Google Cloud team understood Wayfair’s technology needs. Crusenberry says, “The Google Cloud team recognized that the speed and the capacity gain was far more important than anything else. Their willingness to let go of the idea that the implementation needed to be done a certain way was very powerful for us. We didn’t worry that the Google Cloud team would try to push us to use a product or make us configure systems a certain way. They really understood what we were trying to accomplish.”

Moving forward, Wayfair plans to take advantage of the flexible tool set they have as a result of their collaboration with Google Cloud. “What we find very valuable about Google Cloud is the availability of a lot of valuable services; things like Cloud Data Loss Prevention API that can be composited together with other tools to build really compelling applications,” says Ethan Dickinson, Associate Director of Data Infrastructure at Wayfair. “We can use them in a typical way, but we also see a lot of value in integrating them with our internal tools.”