Michael McBride
Founder & CEO, Atelier Commerce
BUILT WITH
Michael McBride
Founder & CEO, Atelier Commerce
BUILT WITH
We didn't rebuild PXG. We layered a personalization stack on top of a store already doing volume, and in 104 days it lifted conversion nearly 25%, revenue per session nearly 28%, and add-to-cart almost 17%, while sessions were down roughly 20% seasonally. Same traffic, more revenue.
We didn't rebuild PXG. We layered a personalization stack on top of a store already doing volume, and in 104 days it lifted conversion nearly 25%, revenue per session nearly 28%, and add-to-cart almost 17%, while sessions were down roughly 20% seasonally. Same traffic, more revenue.
The Challenge
PXG was already a mature Shopify storefront with real volume, a loyal audience, and a high-consideration category. What it did not have was a personalization layer working on its behalf.
Every session was treated the same. Homepage, category, PDP, cart, and 404 all served static content. Search worked, but only after a shopper knew what to type. Bundle merchandising was not in play. Behavioral segmentation, none. On a store doing real volume every quarter, the recommendation engine was not just idle. It did not exist. The brief was clear: stand up Nosto end-to-end, get campaigns serving quickly, and prove the lift on the same store, same season, same catalog. Instrumentation had to go in without breaking anything. A live Shopify store at PXG's volume does not have a maintenance window. Tracking needed to fire cleanly from day one and not affect page performance. Coverage had to be near-total. Recommendations only pay back when they run everywhere shoppers actually go: homepage, category, PDP, cart, search results, 404. Half-measured coverage kills the story before it starts. Merchandising had to be real. A full bundle library, product sets curated, header templates written, category logic mapped. Not a template, a merchandising plan. The measurement window had to be honest. We built two matched 104-day windows: one before campaigns served (tracking only), one after they served. Same store, same season overlap, no cherry-picking.
The Challenge
PXG was already a mature Shopify storefront with real volume, a loyal audience, and a high-consideration category. What it did not have was a personalization layer working on its behalf.
Every session was treated the same. Homepage, category, PDP, cart, and 404 all served static content. Search worked, but only after a shopper knew what to type. Bundle merchandising was not in play. Behavioral segmentation, none. On a store doing real volume every quarter, the recommendation engine was not just idle. It did not exist. The brief was clear: stand up Nosto end-to-end, get campaigns serving quickly, and prove the lift on the same store, same season, same catalog. Instrumentation had to go in without breaking anything. A live Shopify store at PXG's volume does not have a maintenance window. Tracking needed to fire cleanly from day one and not affect page performance. Coverage had to be near-total. Recommendations only pay back when they run everywhere shoppers actually go: homepage, category, PDP, cart, search results, 404. Half-measured coverage kills the story before it starts. Merchandising had to be real. A full bundle library, product sets curated, header templates written, category logic mapped. Not a template, a merchandising plan. The measurement window had to be honest. We built two matched 104-day windows: one before campaigns served (tracking only), one after they served. Same store, same season overlap, no cherry-picking.
The System
We deployed Nosto in phases, activated in a single go-live, and measured against a matched pre-window on the same store.
Deployment and instrumentation. Nosto site-metrics tracking went live in early summer. For the following ~14 weeks, the tag was firing on every session, collecting behavioral data, product-view patterns, and cart signals, with no campaign impressions yet. That preceding period became our clean control window. Campaign build-out. Between mid-August and mid-September, we configured a full campaign library spanning recommendations, dynamic bundles, and post-purchase surfaces. Nearly every available recommendation placement was activated (43 of 44): homepage, category page, product page, shopping cart, search results, 404 page, and general layouts. Dynamic bundle merchandising. Over one hundred bundles were built in the week before launch. Themed templates covering bottoms, tops, outerwear, and accessories, with multiple products per bundle. This was a merchandising investment, not a plugin toggle. Autocomplete / search suggestion dropdown activation. The client kept Shopify native search for full search results (existing rankings and merchandising logic remained intact), and Nosto's autocomplete/search-suggestion layer was activated on top, with a full library of merchandising rules configured to shape the dropdown experience. Segmentation and audience infrastructure. A full library of segments was configured in Audience Insights & Builder, including affinity and behavioral cohorts, laying the groundwork for future personalized campaigns. Go-live and measurement. Campaigns began serving in mid-September. Steady-state impression volume was reached within about five days. We measured the fully-active window (104 days) against the immediately preceding tracking-only window (105 days). Nosto's own analytics generated the comparison; no external attribution model, no synthetic uplift.
The System
We deployed Nosto in phases, activated in a single go-live, and measured against a matched pre-window on the same store.
Deployment and instrumentation. Nosto site-metrics tracking went live in early summer. For the following ~14 weeks, the tag was firing on every session, collecting behavioral data, product-view patterns, and cart signals, with no campaign impressions yet. That preceding period became our clean control window. Campaign build-out. Between mid-August and mid-September, we configured a full campaign library spanning recommendations, dynamic bundles, and post-purchase surfaces. Nearly every available recommendation placement was activated (43 of 44): homepage, category page, product page, shopping cart, search results, 404 page, and general layouts. Dynamic bundle merchandising. Over one hundred bundles were built in the week before launch. Themed templates covering bottoms, tops, outerwear, and accessories, with multiple products per bundle. This was a merchandising investment, not a plugin toggle. Autocomplete / search suggestion dropdown activation. The client kept Shopify native search for full search results (existing rankings and merchandising logic remained intact), and Nosto's autocomplete/search-suggestion layer was activated on top, with a full library of merchandising rules configured to shape the dropdown experience. Segmentation and audience infrastructure. A full library of segments was configured in Audience Insights & Builder, including affinity and behavioral cohorts, laying the groundwork for future personalized campaigns. Go-live and measurement. Campaigns began serving in mid-September. Steady-state impression volume was reached within about five days. We measured the fully-active window (104 days) against the immediately preceding tracking-only window (105 days). Nosto's own analytics generated the comparison; no external attribution model, no synthetic uplift.
Success Stories
Scarlett Gasque: The Open Was Never the Problem
A Klaviyo program that looked healthy on every surface metric was quietly leaving revenue on the table. We found the leak in a single number and fixed it in 90 days, without a rebuild. Sessions fell 32%. Revenue rose 78%. The brand didn't get more visitors. It got better ones, and converted them.
Scarlett Gasque: The Open Was Never the Problem
A Klaviyo program that looked healthy on every surface metric was quietly leaving revenue on the table. We found the leak in a single number and fixed it in 90 days, without a rebuild. Sessions fell 32%. Revenue rose 78%. The brand didn't get more visitors. It got better ones, and converted them.
Social Media Marketing
Alicia Adams Alpaca: 90% First Impressions from Organic Social
alicia adams alpaca runs organic social as top-of-funnel discovery, not bottom-funnel conversion. Nearly 90% of visitors it sends are first-timers, the brand's highest new-visitor rate of any channel, arriving curious after seeing a styled throw or editorial content, ready to look closer.
Social Media Marketing
Alicia Adams Alpaca: 90% First Impressions from Organic Social
alicia adams alpaca runs organic social as top-of-funnel discovery, not bottom-funnel conversion. Nearly 90% of visitors it sends are first-timers, the brand's highest new-visitor rate of any channel, arriving curious after seeing a styled throw or editorial content, ready to look closer.
Success Stories
Scarlett Gasque: The Open Was Never the Problem
A Klaviyo program that looked healthy on every surface metric was quietly leaving revenue on the table. We found the leak in a single number and fixed it in 90 days, without a rebuild. Sessions fell 32%. Revenue rose 78%. The brand didn't get more visitors. It got better ones, and converted them.
Social Media Marketing
Alicia Adams Alpaca: 90% First Impressions from Organic Social
alicia adams alpaca runs organic social as top-of-funnel discovery, not bottom-funnel conversion. Nearly 90% of visitors it sends are first-timers, the brand's highest new-visitor rate of any channel, arriving curious after seeing a styled throw or editorial content, ready to look closer.
