Agentforce Commerce: The ROI Case for AI Agents in Retail

AI-driven traffic to retail sites surged 693% year-over-year during the 2025 holiday season, according to Adobe Analytics data covering more than 1 trillion visits to U.S. retail sites. That is not a projection. It already happened. And the retailers who had autonomous agents in place when that traffic arrived converted it at rates 31% higher than other traffic sources.

During the 2025 holiday season, AI agents and generative AI tools influenced more than 20% of all online retail sales globally, according to Salesforce. AI-referred shoppers spent 45% more time on-site, viewed 13% more pages per visit, and generated 254% higher revenue per visit than non-AI traffic sources.

The question retail leaders are now asking is not whether AI agents work in commerce. The proof is in. The question is whether their organization will be positioned to capture the next wave or spend another cycle watching competitors do it first.

This article makes the business case for Agentforce Commerce using real data from real retailers. Not vendor projections. Named brands, hard numbers, and an honest look at where implementation complexity can stall ROI if the foundation isn’t right.

Key Takeaways

• AI-driven traffic to retail sites grew 693% YoY in the 2025 holiday season. AI-referred shoppers convert at rates 31% higher than other sources.
• Agentforce Commerce is an autonomous AI agent layer built on unified customer data, not a chatbot.
• R.M. Williams saw 34% revenue growth and a 20% conversion lift. Wiley hit 213% ROI.
• The highest-impact use cases: guided shopping, WISMO automation, and personalized recommendations.
• Disconnected data is the primary ROI blocker. Fix the foundation before deploying agents.
• 90% of sellers plan to use AI agents by 2027. The early mover window is 12-18 months, not 5 years.

What Is Agentforce Commerce?

Salesforce Agentforce Commerce is the AI agent layer built natively on Data Cloud and Salesforce Commerce Cloud. Formerly known to many retailers as Salesforce Commerce Cloud AI, the platform has evolved well beyond its roots as a recommendation engine. The architecture matters here: this is not a chatbot bolted onto an existing platform. Agents operate directly on unified customer data, pulling from CRM records, purchase history, behavioral signals, and real-time inventory simultaneously, and act autonomously across the full commerce journey without waiting for a human to trigger each step.

That distinction separates Agentforce Commerce from the wave of AI chat tools retailers experimented with over the past two years. Standard chatbots follow scripts. Agentforce agents reason, make decisions, execute multi-step tasks, and escalate to human agents only when genuinely needed.

The platform’s core capabilities span the entire purchase lifecycle:

• Guided Shopping: Conversational agents that help customers navigate high-consideration purchases, ask qualifying questions, and surface the right product at the right moment
• Personalized Recommendations: AI-driven product suggestions built on unified customer data, not generic collaborative filtering
• Order Management and Post-Purchase: Autonomous handling of WISMO queries, returns initiation, and order modifications without human intervention
• Promotions and Pricing: Dynamic offer deployment based on real-time behavioral signals and customer segment data
• Cross-Channel Continuity: Agents operate consistently across web, mobile, and third-party AI platforms via the Agentic Commerce Protocol (ACP) and Agent Payments Protocol (AP2)

“Consumers are reviewing more content pre-purchase, making Guided Shopping and AI personalization key to closing sales.” — Nitin Mangtani, SVP/GM, Agentforce Commerce, Salesforce

The unified data foundation is the differentiator. Agents are only as effective as the data they act on. For retailers already running on Salesforce Commerce Cloud, that foundation is largely in place. For those who aren’t, it becomes the first implementation conversation. To understand what Salesforce Agentforce is at the platform level, including its architecture and agent builder, that context is worth establishing before evaluating the commerce-specific layer.

Agentforce for Retail: What Early Adopters Are Seeing

The most important thing to know about Agentforce for retail is that the ROI is no longer theoretical. Several retailers have published results, and the numbers are specific enough to anchor a business case.

R.M. Williams: The Retail Proof Point

R.M. Williams is the clearest retail case study available. The Australian heritage footwear brand deployed Agentforce to power guided shopping experiences and personalized product recommendations. The outcome: 34% revenue growth and a 20% conversion rate increase. For a brand selling considered, premium products where the path to purchase involves real decision-making, those numbers reflect exactly what guided shopping agents are designed to do. They reduce friction at the moment of intent.

Pandora: Customer Experience as a Revenue Driver

Pandora’s deployment focused on automating high-volume customer inquiries and deploying AI-powered recommendations at scale. The result was a 10% NPS lift, a customer satisfaction gain that directly correlates with repeat purchase rates and lifetime value in jewelry retail. As David Walmsley, Pandora’s Chief Digital and Technology Officer, put it: “We are redefining shopping with true dialogue via trusted AI and unified data.”

The Broader Productivity Case

The brand-level results align with wider data from Salesforce’s State of Sales report. The productivity gap between AI-adopting and non-adopting teams is widening fast.

83% of sales teams using AI reported revenue growth in the past year, compared to 66% of teams without AI. Top-performing sellers are 1.7 times more likely to use AI agents than their lower-performing peers.

Nearly 90% of respondents plan to adopt AI agents by 2027. And 94% of sales leaders who have already adopted them agree they are essential to growth.

Wiley’s 213% ROI is the highest in the published data set. It also signals that the Agentforce model scales beyond pure retail into commerce-adjacent sectors, a useful data point for retailers with complex B2B or wholesale channels alongside their consumer business.

Key Use Cases Driving Results

Understanding where Agentforce Commerce generates the most impact helps retailers prioritize deployment. The use cases below are ranked by the strength of available evidence, not by platform positioning.

Guided Shopping for High-Consideration Purchases

When customers face real trade-offs, fit, compatibility, personalization, price, a guided shopping agent outperforms a static product page. It asks qualifying questions, narrows the product set, and surfaces the right recommendation with context.

This is the mechanism behind R.M. Williams’ 20% conversion lift. The agent reduces the cognitive load that causes abandonment on high-ticket items.

63% of Gen Z consumers say they are interested in using AI agents to help make purchases, according to Salesforce. That cohort is entering peak spending years. Retailers who build guided shopping capability now are building for the dominant consumer demographic of the next decade.

Post-Purchase and WISMO Automation

“Where is my order?” queries are the single highest-volume contact reason in most retail service operations. They are also entirely automatable. According to Salesforce, a single WISMO request costs a business around $12 to resolve manually. Agentforce post-purchase agents handle these queries, returns initiation, and order modification requests without human involvement, freeing service teams for genuinely complex cases.

Shoe Carnival projects its Agentforce agent will handle 40% of inbound call volume autonomously. That is not a marginal efficiency gain. At scale, it represents a structural reduction in service cost that compounds as order volume grows.

Personalized Recommendations at Scale

During the 2025 holiday season, AI agents and generative AI tools influenced more than 20% of all online retail sales globally, according to Salesforce. The scale reflects what happens when recommendation logic runs on unified customer data rather than session-level signals. Agents that know a customer’s full purchase history, browsing behavior, and stated preferences produce recommendations that convert. Agents working from anonymous session data produce noise.

AI-Driven Promotions

Agentforce agents can deploy personalized promotions in real time based on behavioral triggers: cart abandonment, browsing patterns, loyalty tier, and predicted churn risk. This moves promotional strategy from batch campaigns to individual-level intervention, which is where how AI drives commerce growth at the margin level becomes most tangible.

The Implementation Reality: Where ROI Can Stall

The ROI data is compelling. The implementation complexity is real. Retailers who go in expecting a plug-and-play deployment tend to underestimate the data work required before agents can perform at the level the case studies describe.

The four most common friction points, based on Salesforce’s own research and published case study data:

• Disconnected data is the primary blocker. Agents act on the data available to them. If customer records, inventory systems, and order management platforms aren’t unified, agents make decisions on incomplete information, producing poor recommendations, failed automations, and eroded trust. Commerce optimization, cleaning product data, connecting systems, and eliminating integration bottlenecks, is a prerequisite, not a parallel workstream.
• Enterprise application integration remains low. Only 27% of enterprise applications are currently integrated, according to Salesforce research. That means most retailers deploying agents are doing so against a fragmented data environment. The gap between what agents can theoretically do and what they actually do in a siloed architecture is significant.
• Agent isolation limits compound value. 50% of deployed agents currently operate in isolation. The full value of Agentforce Commerce comes from agents working in orchestration: guided shopping agents handing off to post-purchase agents, service agents surfacing data to marketing agents. Isolated deployments capture a fraction of the available ROI.
• Shopper adoption is not guaranteed. Some consumers still prefer completing purchases through traditional web or mobile checkout. Conversion from AI-assisted discovery to completed order needs optimization, particularly in the early phases of deployment.

“Traffic from tools like ChatGPT, Gemini, and Perplexity is growing rapidly. More importantly, it’s high-quality traffic that converts. For retailers and brands, this signals a lasting shift in how consumers discover and evaluate products.” — Vivek Pandya, Lead Analyst, Adobe Digital Insights

None of these challenges are blockers. They are sequencing decisions. Retailers who address the data foundation first deploy agents that perform closer to the published benchmarks from day one.

Why the Window to Act on Agentforce Commerce Cloud Is Narrowing

The adoption curve for AI agents in commerce is not moving on a five-year timeline. It is moving on an 18-month one.

54% of sellers are already using AI agents. 90% plan to by 2027. Agentforce’s ARR grew 114% year-over-year to $1.4 billion in Q3 FY2026, with account growth of 70% quarter-over-quarter. 76 of North America’s top 2,000 retailers already run on Salesforce Commerce Cloud, meaning the integration path to Agentforce Commerce exists for a substantial portion of the market right now.

The 2025 holiday season generated $257.8 billion in online sales, with AI-referred traffic converting at rates 31% higher than other sources. The 2026 season will be larger. Retailers who have agents in place, trained on real customer data, and integrated across their commerce stack before that window opens will capture a disproportionate share of it. Those deploying in Q4 will be optimizing in real time against competitors who started a year earlier.

Three steps retailers should take now:

1. Audit your data foundation. Identify where customer data, inventory, and order management systems are disconnected. The data gaps you find are the gaps that will limit agent performance.
2. Identify one high-impact use case to deploy first. Guided shopping or WISMO automation are the fastest paths to measurable ROI. Start there, prove the model, then expand.
3. Work with a Salesforce implementation partner who has live Agentforce deployments, not just platform certifications. The sequencing decisions that determine whether you hit 34% revenue growth or 10% require experience, not just credentials.

The ROI case for Agentforce Commerce is no longer a forward-looking argument. It is a present-tense one. The retailers in the case studies above acted. Working with an experienced Salesforce implementation partner is the fastest way to close the gap between where your data foundation is today and where it needs to be before peak season.

Frequently Asked Questions

What Is Agentforce Commerce and How Does It Work?

Agentforce Commerce is Salesforce’s AI agent layer built natively on Data Cloud and Commerce Cloud. Unlike chatbots, it operates autonomously across the full commerce journey, from guided shopping and personalized recommendations to order management and post-purchase service, using unified customer data to act in real time.

What ROI Can Retailers Expect from Agentforce Commerce?

Early results are strong. R.M. Williams saw 34% revenue growth and a 20% conversion rate lift. Pandora achieved a 10% NPS improvement. Wiley reported 213% ROI. Across the board, 83% of sales teams using AI reported revenue growth, compared to 66% of teams without AI.

How Does Agentforce Commerce Differ from a Standard AI Chatbot?

Agentforce Commerce agents act autonomously on unified customer data, pulling from CRM, purchase history, and behavioral signals simultaneously. Standard chatbots follow scripts. Agentforce agents reason, make decisions, execute multi-step tasks, and hand off to human agents only when genuinely needed.

What Are the Main Use Cases for Agentforce Commerce in Retail?

The highest-impact use cases are guided shopping for high-consideration purchases, post-purchase and WISMO query automation, personalized product recommendations at scale, and AI-driven promotions. Shoe Carnival, for example, projects its Agentforce agent will handle 40% of inbound call volume.

What Are the Biggest Implementation Challenges for Agentforce Commerce?

Disconnected retail data is the primary blocker. Agents are only as effective as the data they act on. Only 27% of enterprise applications are currently integrated, and 50% of deployed agents operate in isolation. Retailers need a unified data foundation before agents