Connected Data Makes AI Actionable
Generated by Klaviyo AI
Origin connects customer, production, and behavioral data so AI can analyze the full customer journey and support objective decisions. The approach helped identify a denim fit issue expected to save $150,000 annually, while humans remain responsible for validating findings and preserving customer-facing authenticity.
- Unified customer context: Connecting data across systems lets signals from returns, reviews, engagement, and purchases inform each other.
- Measurable AI impact: AI linked enzymes in a denim wash to inconsistent shrinkage, leading to a fix expected to save $150,000 per year.
- Human validation: Teams still need to ask useful questions, verify AI recommendations, and decide how to act on them.
Justin Parker, director of ecommerce at American-made apparel company Origin, operates on one rule: Control the whole chain, not just pieces of it. He applies that philosophy to both growing the denim brand and building a connected customer data strategy.
Origin makes nearly everything in-house, in the US, from raw materials to the factory floor. The brand also pulls their customer data from one unified customer profile, rather than piecing together separate tools to aggregate data from email, service, analytics, carriers, warehouses, and production.
In our latest episode of Klaviyo Originals, Parker shares how Origin handles customer data and AI, how the brand preserves authenticity, and how he turned a hunch about ill-fitting denim into a fix worth an estimated $150,000 a year with the help of AI.
Origin’s philosophy on owning the entire customer journey
Origin owns its supply chain, materials, and manufacturing for one rationale.
"The reason why we have everything in-house and why we make everything in America is control," Parker said. "We have control over our supply chain. We can show up better for our customers because we can plan for the things that we're making."
That same logic extends to customer data.
"With one system, your data is centrally located, with many different integrations feeding into it," Parker said. "When you aggregate the data, it's all coming in the same context with the same structure and schema."
That structure means a signal from one part of the customer journey carries into the next. A return connects to an open rate, a review connects to a piece of feedback, and feedback connects to what a customer buys next.
"No matter what the thing is that we're trying to understand, whether it's returns, open rates, reviews or feedback or whatever, it's all connected," Parker said. "We can leverage the context that's created from one part of the customer journey to another part of the customer journey."
Knowing what a customer is reacting to and what's actually resonating with them helps ads, email, and landing pages land for the right segments, keeping brands top of mind for customers.
"Having that understanding of your customer helps you drive better segmentation. It really is the driver to make sure that your brand is showing up for the customers that your brand resonates with,” Parker said.
But collecting that data is just the first step. Next, brands need to act on what they know.
"Storing customer data is useless unless you're actually doing something with it,” said Parker. “You should be able to act on that data from every angle that the customer interacts with you: on your site, through email, through ads, through how your products are developed, through how your business operates as a whole, and through how your customer service team interacts with customers."
How a hunch became a $150,000 fix thanks to AI analysis
Origin's connected data setup got tested when the customer service team noticed something that didn't add up.
"Customer service had a hunch that there was a particular wash of our denim that wasn't fitting quite right. It wasn't consistent with the other washes," Parker said.
Instead of guessing, Parker gave an AI tool the whole picture, including return records, the tech pack that specifies how the denim is washed, which fabrics and threads it uses, and the chemicals in the wash process.
"I ended up just asking AI and giving it the situation," Parker said. "Here's our hunch, either validate it or invalidate it, and let us know if this is something we should chase or not."
The AI found that the enzymes used in the light wash were shrinking the denim differently than the other washes, even though the pattern was identical. It went further, recommending Origin tighten specific measurement tolerances from a half inch to a quarter inch.
Parker brought the finding to the product team, and the fix worked. Origin now expects to save an estimated $150,000 a year in returns, exchanges, and shipping costs tied to that one wash.
"What I like about these types of explorations is that they're so objective. It's not 'I feel' or 'this seems to me,'" Parker said. "It's, ‘this is what the data is showing.’ Then it's up to the individual business units to go and chase it down and see if it's valid and if it makes sense."
Where AI helps, and where Origin keeps humans in charge
Origin uses AI to help with data analysis, but stops at the words and images customers actually see to preserve authenticity.
"We owe it to the customer to be genuine and to put our best foot forward," Parker said. "We're not a fast-fashion brand. We're a brand that's like the new American heritage. Everything is built by people that are paid a living wage in America, so it does command a higher price. We owe it to our customers to put our best foot forward and to put copy and content out there that is worth the cost of the item."
AI still touches production in small ways, like swapping a color on a product photo. The story, the message, and the angle, however, stay off-limits.
"Telling the story or coming up with the angle that we want to help the customer know how the product is going to feel, those things are Origin all the way," Parker said. "And I don't think that's ever going to change. The route that we take to get there might change, just like we're not using film cameras anymore, we're using digital cameras. But the story, the people, the products, the process, that's always going to be authentic to Origin."
In other areas of the business, AI runs with more independence. Origin's paid media budgeting, for example, is fully automated within guardrails the team set in advance.
"We worked really hard to set the guardrails, to set the different criteria for budgeting," Parker said. "We obviously monitor it, but on a day-to-day basis, we don't actually have anybody that's going in and changing budgets. That's happening through AI."
Here’s how Parker’s team approaches AI in marketing:
- Data analysis: AI sorts through more returns, production, and behavioral data than a person can track alone, and provides recommendations a seasoned team can act on.
- Paid media budgeting: AI bids and adjusts budgets inside guardrails the team sets in advance, with light human monitoring. The person who used to make those calls day to day now spends that time on direct mail, the email calendar, and taking on new initiatives the team hadn't been able to get to before.
- Customer-facing copy and creative: The words and images customers see stay human, because authenticity is core to the brand.
Origin’s advice for newer brands to collect and connect customer data
Origin has been collecting data for two decades, so they have a pretty robust picture of their customer. Parker's advice for brands starting from a much smaller stage is:
- Send a short survey. Ask your list 15 direct questions instead of guessing at what customers want, collecting first-party data. What kind of vehicle do they drive? Are they married? How old are they? The answers to these questions can help you get to know your customer better.
- Connect your systems. Make sure the tools you already have, including email marketing, customer service, website analytics, and even warehouse, production, and carrier data can share information before adding anything new.
- Step away from the dashboard. Set the data aside occasionally and ask what would matter to you as a customer.
He's found customers are often more willing to answer questions than brands expect. "If you have a brand that has any sort of following, your customers are going to be so willing to help you out," Parker said. "We're constantly amazed by the number of results we get when we just flat out ask our customers questions."
After surveying your customers and connecting your systems, he recommends stepping away from the numbers and thinking like a customer.
"Where the wins are is actually disconnecting from all the data and putting yourself in the customer's shoes, and asking earnest questions," he said. "If I was a customer, what matters to me when I'm shopping with this brand? What things would stand out to me if I were considering this brand?"
AI is only as good as the people using it
Connected data makes AI worth using, but people still need to decide what to do with what it finds.
"It's an incredible tool, but it's only as good as the context that you feed it," Parker said. "And frankly, it's only as good as the people that are utilizing it."
For Origin, that meant a customer service hunch became a $150,000 fix only because someone was curious enough to ask the question, and someone else on the product team knew the numbers well enough to check the AI's answer against reality.
Connecting the data was the first step. Asking the right question was the next.
Related content
