How Lumens Cut Manual Merchandising from Hours to Minutes — And Boosted Search Revenue
The business serves both design professionals and design aficionados, helping them create spaces that express their authentic style.
For a retailer like Lumens, product discovery matters. Shoppers often arrive with a specific style, room, or design goal in mind. So, they need to translate this intent into an efficient product discovery experience that supports the brand’s merchandising priorities.
The business serves both design professionals and design aficionados, helping them create spaces that express their authentic style.
For a retailer like Lumens, product discovery matters. Shoppers often arrive with a specific style, room, or design goal in mind. So, they need to translate this intent into an efficient product discovery experience that supports the brand’s merchandising priorities.
Search results often felt off because products that didn’t fit the query would appear high on the page. So, the team had to step in with manual rules, filters, and search merchandising to make results look acceptable. Over time, that created a large operational burden.
That burden showed up in other areas, too. Facet and attribute management required extra manual effort, including updating display names and bucketing values that should have been handled more cleanly by the platform. With a small team, those tasks took time away from higher-value work.
“[A lot of our products] have certain attributes, like a style attribute. Some products have multiple data points within that same attribute (for example, Scandinavian or mid-century). In [our legacy system], I had to manually update the display name for all of those and bucket them."
- Rachel Bews, Site Optimization Manager, Lumens
At the same time, Lumens was managing product discovery across separate tools. Search, Browse, and Recommendations lived in different systems, which made it harder to get a shared view of shopper behavior and harder to let learning in one part of the site improve another.
Search results often felt off because products that didn’t fit the query would appear high on the page. So, the team had to step in with manual rules, filters, and search merchandising to make results look acceptable. Over time, that created a large operational burden.
That burden showed up in other areas, too. Facet and attribute management required extra manual effort, including updating display names and bucketing values that should have been handled more cleanly by the platform. With a small team, those tasks took time away from higher-value work.
“[A lot of our products] have certain attributes, like a style attribute. Some products have multiple data points within that same attribute (for example, Scandinavian or mid-century). In [our legacy system], I had to manually update the display name for all of those and bucket them."
- Rachel Bews, Site Optimization Manager, Lumens
At the same time, Lumens was managing product discovery across separate tools. Search, Browse, and Recommendations lived in different systems, which made it harder to get a shared view of shopper behavior and harder to let learning in one part of the site improve another.
Constructor stood out during the evaluation for a specific reason: the team saw a path to doing less manual work without losing control.
During the product demo, what clicked for Melissa Godes, Head of Ecommerce at Lumens, was how Constructor approached search terms — not just matching keywords, but learning what shoppers actually mean when they type. That logic meant the tool could improve on its own, rather than relying on the team to keep feeding it synonyms and one-off rules.
“We wanted to be hands-off and wanted the tool to learn without having to constantly add more synonyms. We feel comfortable not doing [constant synonym additions] with Constructor."
- Melissa Godes, Head of Ecommerce, Lumens
Constructor stood out during the evaluation for a specific reason: the team saw a path to doing less manual work without losing control.
During the product demo, what clicked for Melissa Godes, Head of Ecommerce at Lumens, was how Constructor approached search terms — not just matching keywords, but learning what shoppers actually mean when they type. That logic meant the tool could improve on its own, rather than relying on the team to keep feeding it synonyms and one-off rules.
“We wanted to be hands-off and wanted the tool to learn without having to constantly add more synonyms. We feel comfortable not doing [constant synonym additions] with Constructor."
- Melissa Godes, Head of Ecommerce, Lumens
.png?width=80&height=80&name=member-certified-24-white%20(1).png)
