Temple & Webster Excels With Server-Side A/B Testing

With a long-standing investment in web optimization and testing, Temple & Webster have refined their approach and embraced optimization site-wide, especially by using Origin Experiments©, SiteSpect’s unique “in the flow” server-side approach, to refine and improve the functionality of their digital experiences beyond look and feel.

Temple & Webster’s Key To Success

Temple & Webster had tested other optimization tools in the past, but felt they fell short when it came to really moving the needle. Specifically, they found tools that focused primarily on client-side look and feel changes didn’t provide the level of sophistication they required, and any server-side offerings fell short in complexity, workflow, or integration capability. Because of SiteSpect’s patented reverse proxy architecture, they were able to start experimenting at the origin without disrupting the existing code or impacting site performance. Mike Henriques, CIO at Temple & Webster says about doing Origin Experiments©, “You simply must be doing this. We haven’t seen a tool that has allowed us to do as much complex and statistically significant testing as we have been able to do with SiteSpect.”

Since implementing SiteSpect in 2018, Temple & Webster consistently has at least four Origin Experiments© running and four client-side experiments running at any given time. Henriques describes SiteSpect as, “Complexity at our end, and simplicity on the SiteSpect side.” SiteSpect integrates completely with the Temple & Webster online presence, meaning the team can test anything and everything related to their site, whether it pertains to an internally built tool or a third-party service.

The Importance of Data-Driven Decisions

We haven’t seen a tool that has allowed us to do as much complex and statistically significant testing as we have been able to do with SiteSpect.

Mike Henriques, CIO, Temple & Webster

Temple & Webster have always provided a customer-focused, data-driven digital experience, and with a strong testing and optimization program, they can understand the impact of any change on any channel. Brodie mentions that, “The most valuable insights aren’t necessarily the obvious big wins or losses. Instead, the most valuable results are those that differ from our hypotheses, common wisdom or accepted UI or UX ‘best’ practices.”

While there are some great standards out there about what type of copy, font, or search algorithm work best, the actual results are incredibly site and context-specific. As Brodie says, “When the data differs from expectations, that’s where you really make data-driven decisions.” He also noted that SiteSpect’s technology provides better data accuracy, more seamless integration and visibility across the entire customer journey compared to what they experience with tag-based vendor tools.

Testing Across the Company

With the measurable successes Temple & Webster have seen with SiteSpect, and the “second to none” support, as Brodie and Henriques noted, testing and optimization has begun to spread throughout the organization. Temple & Webster merchandising and marketing teams have taken some of the A/B testing into their own hands, meaning they can more efficiently validate and improve their work. Henriques says that SiteSpect is “enabling more of the business to be self-sufficient in testing some of their own hypotheses, which they couldn’t do before. And that’s a hugely positive driver of our culture of data-driven innovation.”

Mobile Page Redesign

Recently, Temple & Webster launched a page redesign for their mobile site. As part of the process, they tested all of their important metrics on the new experience. After that initial round of testing, they discovered that the results were inconclusive, and it wasn’t clear which components of the site were helping or hurting conversions. This type of result is fairly common in testing if a lot of changes appear in one variation, and it’s a stepping stone to further inquiry.

From this initial result, the team worked with SiteSpect to break down the new mobile page into smaller pieces, creating new variation groups for each change. Temple & Webster then rolled out each change on its own to learn the impact each element had both individually and in conjunction with other elements via multivariate testing. By isolating each change, the originally flat result took on a whole new shape. It turned out that just one element caused a negative turn in conversions, and that without that single element the redesign increased conversions significantly.

With this knowledge, Temple & Webster not only fine-tuned their mobile page redesign but also found a potentially problematic site component that they could move forward to test and then address site and device-wide.

Conclusion

Since implementing SiteSpect, Temple & Webster has taken advantage of the breadth of capabilities it offers — using Origin Experiments© extensively, putting optimization in the hands of many departments, and collecting accurate and thorough data about their user experience. Brodie and Henriques emphasize the partnership with SiteSpect consultants as key to their success.