Last week we hosted the webinar edition of an event that we hosted during NRF 2026. This webinar featured leaders from Marketplacer, Spur, Fluent Commerce and Versori (of course) to dissect how each company joins together to provide a combined marketplace AI solution, as well as how each company is using AI to combat specific marketplace challenges. Here are the key takeaways...

AI in Marketplaces: From Experimentation to Infrastructure

Marketplaces add operational complexity, fast. The model selected at the start shapes everything that follows: seller onboarding, catalogue quality, order routing, returns, and the ability to protect brand promises. AI adoption in this environment works when foundations exist: visibility, connected systems, and measurable outcomes.

Marketplace models: start with strategy, not labels

“Marketplace” covers multiple operating models, and the right choice depends on the goal.

Open marketplaces

Open marketplaces focus on scale:

• Broad seller base
• Fast SKU growth
• Higher governance requirements
• More operational complexity
• Brand risk that must be manageable
• Works best when traffic exists to support the model

Curated (vertical) marketplaces

Curated marketplaces focus on control and brand:

• Fewer, aligned sellers
• Stronger control over the offer
• More curation
• Value delivered to an existing customer base

Hybrid approaches

A common direction is a hybrid:

• A curated core
• A controlled long tail

The best-fit model is the one that strengthens the brand and scales with operational maturity, rather than maximizing seller count and product volume.

Why marketplaces change buying behaviour

Marketplaces shift a simple retail flow into many possible paths:

• Multiple ways to order
• Multiple ways to process returns
• More sources of failure across sellers, systems, and fulfillment paths

This is why order management and quality assurance become central topics early.

Intelligent order orchestration in OMS: what it means

Rules-based processing remains necessary for deterministic flows. Intelligent orchestration adds a phased approach that uses data to improve decisions over time.

The operating framework: measure, learn, act

1. Measure
   • Visibility into order sourcing performance
   • Impact on margin
   • Ability to deliver on time and in full
2. Learn
   • Insights into what is working vs. not working
   • A/B testing across sourcing profiles
3. Act
   • Human-in-the-loop changes where appropriate
   • Autonomous actions where appropriate

Where AI fits

AI supports optimization and proactive operations at a scale humans cannot match, including:

• Recommendations that balance competing goals (fast delivery and high margin)
• Proactive identification of orders at risk for on-time/in-full performance
• Ongoing monitoring across the full order set
• Customer communication when risk is detected

This progression reduces risk compared to immediate “hands-off” automation.

AI-driven QA for marketplaces: measuring what is out of control

Marketplaces introduce many more failure points than a single-brand commerce site because product managers do not control every seller’s content and execution. AI agents can simulate customers moving through the marketplace to detect issues that are otherwise hard to cover.

What gets measured

• Broken images
• Incorrect PDP descriptions
• Video loading failures
• Localization and translation issues
• Checkout and payment errors
• Support tickets tied to failed order flows

Measurement becomes the proof point because testing every item and path manually is not feasible.

Why automation matters

Automation enables coverage at scale:

• The equivalent of many human testers operating continuously
• Broader detection across the marketplace catalog and flows

How rollout typically starts

Adoption often begins with a narrow proof of concept:

• A small set of SKUs
• A single pain point (localization, PDP media failures, checkout errors) Then expands as issues are mapped to bugs and tied to conversion impact.

Seller-side tooling: the lever for growth and competition

Seller tooling is a major lever for marketplace growth. The focus has shifted from seller volume to seller quality.

Why seller experience matters

Professional marketplace sellers often sell across many marketplaces and use channel managers. A substandard seller experience pushes sellers elsewhere.

Defining a “good seller”

Seller due diligence includes:

• Desktop research on existing marketplace presence
• Ratings and reviews
• Direct-to-consumer channel maturity
• Product information quality
• Clear returns and delivery proposition
• Signals of modern commerce operations

Operationalizing seller vetting with AI

Seller evaluation data is increasingly available via APIs and internet search. AI enables this work at scale by handling:

• Structured checks (fields, completeness, compliance)
• Unstructured inputs (comments, reviews, text-heavy seller information)

This can automate most of the seller evaluation stage while increasing the amount of assessment performed.

Data readiness: the foundation for AI in marketplace operations

AI does not work well across data silos. Operational gains require visibility across systems, including tightly integrated components such as carrier integrations and fulfillment data.

Two consistent requirements emerge:

• Connected systems to unlock data between platforms
• Measurement tied to outcomes, not feature adoption

Engineering and cybersecurity teams often evaluate AI from security and runtime perspectives, while operational leaders focus on outcomes.

ROI: how AI initiatives get justified

AI adoption is guided by clear ROI math and fast proof of value.

Common ROI anchors include:

• Doing more with fewer resources
• Releasing faster by removing testing bottlenecks
• Reducing lost orders caused by checkout or payment errors
• Improving key operational metrics through measurable changes

The trend through 2026–2027: composable marketplace architecture

A growing preference is emerging against marketplace monoliths. Many operators already have established systems (commerce platform, OMS, PIM, call center). The direction is toward:

• Marketplace platforms that slot into an existing ecosystem
• Composable architectures that allow specialist solutions to be added to deliver outcomes

AI adoption then becomes part of a broader operational design: connected foundations, measurable improvements, and targeted automation where it is safe and valuable.