WATCH A REPLAY OF THE WEBINAR HERE
Key takeaways:
- Enterprise omnichannel execution typically breaks at the inventory and fulfillment layer, not the storefront, according to ESW and Lazer Technologies executives.
- Only about 6% of consumers are willing to purchase directly through social media despite it being the top product discovery channel, creating a widening trust gap brands must close.
- Brands that point AI at fragmented backend systems risk accelerating dysfunction rather than improving performance, making clean product data and unified orchestration the necessary first step.
For retail brands experiencing rapid growth, success itself can become the problem. The more channels a brand activates, including DTC websites, marketplaces, social commerce and emerging AI-powered surfaces, the more likely execution is to quietly break behind a perfectly functional storefront.
That was the central argument made by Frank Kouretas, Chief Product Officer at ESW, and Serge El Hachem, who oversees solutions engineering and technical architecture at Lazer Technologies, Shopify’s enterprise implementation partner, during a June 2026 Retail TouchPoints webinar titled “Fragmented Commerce: Where Enterprise Growth Breaks Down.”
The Storefront Looks Fine. The Back End Doesn’t.
Kouretas and El Hachem opened with a diagnosis: the front end of most enterprise commerce experiences is not where things go wrong.
“It’s usually everything that’s behind it,” El Hachem said. “Different fulfillment rules, different inventory views for each channel, nothing actually unifying them properly. The exact same order can behave completely differently depending on where it came from.”
He described a client situation that illustrated the problem. A global apparel brand selling across DTC and multiple marketplaces had strong traffic and healthy demand numbers but couldn’t explain a persistent pattern of cancellations and refunds, especially around peak periods.
The root cause: each channel was reading inventory from a different source on a different refresh schedule. Units would sell simultaneously on a marketplace and the DTC site, and an order would quietly cancel a day later due to stockout, with the team only learning about it through refund queues and support tickets, never in real time.
“What we had to do was give them a shared view of inventory and write every order against that single source of truth,” El Hachem said. “That essentially fixed it for them.”
Discovery Has Fragmented, But Trust Hasn’t Followed
Beyond fulfillment, the panelists identified product discovery as a second major fracture point, particularly as AI-powered surfaces join search, social and marketplaces in the mix.
Social remains the dominant product discovery channel globally. Roughly half of all consumers use it to find products, with that figure climbing to over 70% for Gen Z, El Hachem said. But AI is beginning to fragment discovery further, and he noted that the shift is uneven: discovery has moved, but purchasing intent hasn’t kept pace.
“Only about 6% of people actually want to buy directly through social, even though it’s the number one product discovery channel,” El Hachem said. “You’ve got this widening gap between where people are discovering your product and where they’re actually willing to hand over their money. That gap is essentially a trust gap.”
Kouretas noted that the trust issue extends well beyond payment. “Even before the payment, how can I trust the information I have about the product? Can I trust that the inventory is actually there?” he said. “There’s definitely some work to do to make it as seamless as the commerce experiences consumers are used to when they buy directly from a brand.”
El Hachem added a data point that underscored the scope of the challenge: only about 3% of consumers are currently comfortable with AI-powered payments. Yet brands that have removed friction through more localized checkout models have seen conversion improvements of around 16%, he said, suggesting the upside is real once trust barriers are resolved.
What Orchestration Actually Means in Practice
The solution the panelists advocated is what they called an orchestration model, a unifying layer that connects disparate demand channels and supply systems without requiring brands to rebuild from scratch each time they expand.
“You define your execution rules once and apply them across every channel and region,” El Hachem said. “When you’re launching a new market or a new surface, it inherits that logic instead of being rebuilt from scratch every single time.”
Kouretas framed the alternatives, full in-house builds or monolithic single-vendor solutions, as both unworkable at scale. Custom integrations per channel drive up cost and complexity indefinitely. A single monolithic platform, meanwhile, eventually hits the limits of its flexibility.
“Any single monolithic solution you buy is going to get blocked at some point,” Kouretas said. “It’s not going to be able to flex to all the needs it has to meet.”
ESW’s own model, he said, functions as that orchestration layer for global commerce, partnering with third-party carriers, 3PLs and other fulfillment providers and mediating across them so brands can sell into new markets without managing the underlying complexity themselves.
El Hachem offered a practical test for whether a brand’s infrastructure is ready: “Adding a marketplace or a social channel should feel closer to configuration than a rebuild. If it feels like a rebuild every time, the backbone isn’t unified yet.”
Where AI Actually Creates Value and Where It Doesn’t
On the question of where AI fits into this picture, both panelists drew a sharp line between AI as an amplifier and AI as an accelerant for existing dysfunction.
“AI creates value in two places,” El Hachem said. “One, as a coordination layer, detecting friction earlier, coordinating decisions across checkout, payment and fulfillment faster than a person could. And second, on the demand side, when your product data is structured well enough for AI to actually understand and recommend you.”
The risk, he said, is pointing AI at a fragmented backend. “All you’ve done is automate the mess itself. It just makes a good operating model better and a broken one worse.”
Kouretas pointed to product data infrastructure as the prerequisite that applies equally to agentic commerce and traditional channels. “A lot of what customers have to solve when they want to prepare for agentic commerce is not specific to AI,” he said. “It’s about exposing data, exposing access to services and exposing them in the format that AI agents can understand. It still goes back to clean data, proper integration and making your services accessible to third parties that need to use them.”
For brands wondering where to start, the answer from both panelists was the same: product data, then inventory visibility, then fulfillment orchestration, in that order, before adding any new surface.
