A great store associate notices things. When you pick up a candle, breathe it in, and put it back, a watchful assistant already knows what to hand you next.
Omni Retail Enterprises wanted that same helpful experience for every online shopper at Pier 1, Dressbarn, and their four sister brands. These brands were built on a rich experience, where taste, preference, and customer intimacy define the customer experience.
Product Genius delivered that same richness for online shoppers, while driving major gains for the business.
At a glance
- Pier 1 revenue per session
- +28.8%mobile, +25.7% desktop vs. control
- Pier 1 ROI
- 414%with a 50.6% EBIT lift on treated traffic
- Time to launch
- 2 weeksfrom first conversation to live A/B test
- Method
- Randomized A/B tests validated with Bayesian analysis
The problem
Millions of shoppers, all of them strangers
Omni owns six of the best-known names in American retail: Pier 1, Dressbarn, Linens 'n Things, Modell's, Stein Mart, and The Franklin Mint. As part of their growth strategy, they have transitioned from brick-and-mortar to online-only selling. The immersive physical retail experience is now entirely digital. The challenge they face is that their loyal, long-time customers all show up as new anonymous website visitors. No cookie or customer ID. No historical data to drive recommendations.
And like most online retailers they were seeing more and more traffic entering directly on a product page, but without strong engagement cues, most of those shoppers would leave in seconds. Bounce rate on product pages was three times the site's normal rate.
Think about your PDPs as your homepage. Scrutinize them as much as the content you're putting on your homepage, because it's just as important, or maybe more important.
Omni's existing tools were no help. Traditional personalization and recommendation engines try to drive by using the rearview mirror. They study what past crowds bought and update in slow batches. Anonymous visitors and new additions to the catalog leave these systems helpless. And none of the legacy personalization tools can act on what the shopper in front of them is doing in-the-moment.
The solution
A Seller's Agent for every shopper
Omni put Product Genius Seller's Agents on its Pier 1 and Dressbarn product pages. Each agent is the digital equivalent of that great store associate, assigned to one shopper for one visit. It watches where you linger, what you scroll past, what you click on, and what makes you stop. Then it reshuffles a feed of products, reviews, and content to fit your interests, while you're still on the page.
The agents run on the Product Genius Large Interaction Model (LIM), which learns 3,000 times faster than standard large language models. It needs only 3 to 10 interactions to read what a shopper wants. No login, no purchase history, no cookie trail needed.
You don't need identity to personalize. The last 30 seconds of behavior beat a stale purchase history if you can act on it in real time.
The “A-HA” moment
Candle people are scent people
Pier 1 had been cross-selling fragrance by format. Buy a candle, see more candles. That's how the catalog was organized, so that's how the site sold.
The AI agents noticed what shoppers were actually interested in: the scent. Someone who loves a candle's fragrance will happily take that same fragrance as a diffuser, a room spray, or an oil. The agent created an online version of “scent bridging”.
Once the agents pulled that thread, more patterns emerged:
- Texture beats taxonomy. A shopper who clicks on rattan wants rattan, in whatever category it lives.
- The add-to-cart pivot. Once something lands in the cart, the shopper is mostly done exploring. Now they want things that go with it, plus a little reassurance they chose well.
No one at Omni had to write a rule, change a landing page, or set up filters for any of this. The agents found the patterns on their own and acted on them, one session at a time.
It was really cool to see a hard, factual data point that says: here's how you should activate this category.
Watch the full talkJeff Haddox on turning Pier 1's product pages into the front doorCommerceNext · Omni Retail EnterprisesLaunch
Live in two weeks, with nothing new to make
Personalization projects usually come with a long IT runway. This one went from first conversation to a live A/B test in two weeks. The agents incorporated the brand’s existing photos, videos, reviews, and social content automatically, so nobody had to shoot or write a thing.
The learning runs both ways, too. Scent bridging began as something the agents did quietly on the site. Now it shapes how Pier 1 merchandises, on the site and beyond it. Omni’s ecommerce team can now focus on the big picture, with the support of insights coming from the on-site agents.
Humans own brand, assortment, and strategy; the AI operates the storefront, with humans in the loop.
Results
An easy decision
The results speak for themselves. Shoppers that had the help of an AI agent had a better experience and generated more revenue. On the “B” branch with Product Genius, Pier 1 saw:
- Revenue per session rose 28.8% on mobile and 25.7% on desktop.
- The program delivered a 414% ROI.
- EBIT rose 50.6%.
Six brands. Twenty-two employees. Millions of shoppers. And now each shopper has someone paying attention and helping them through a purchase journey.
This is really different from anything we've seen in the market. Even if I had a team of a thousand people, we could not do this in real time. That's why we partnered with Product Genius.
All reported metrics are derived from randomized A/B testing on live production traffic (pier1.com and dressbarn.com), segmented by device. Performance was validated using Bayesian analysis, confirming statistical significance — for example, a 96.9% probability of lift on Dressbarn mobile conversion.
