Dreamforce 2026: Commerce Changes That Matter for Retail Leaders

01 / EXECUTIVE VERDICT

My answer in one page

Salesforce has made the storefront more agent-ready, but the ecommerce leadership decision is still about sequencing, data quality, conversion risk and ownership.

What I would do

1. Protect the revenue engine

   Freeze broad architectural change before peak. Preserve price, promotion, inventory, variant, cart and checkout truth across every AI experience.

2. Pilot Agentic Commerce Search first

   It addresses a measurable customer problem, can run against an existing catalog and has a cleaner test design than a full agent rollout. Start with a defined use case and a measurable business outcome.

3. Treat Carter as a service journey, not a chatbot

   Start with bounded use cases such as comparison, fit, compatibility or guided discovery. Product Comparisons and configurable components are available now; richer page awareness, upsell and authenticated context are scheduled for October.

4. Make AI channels a feed-and-controls program

   ChatGPT and Google/Gemini discovery are new acquisition surfaces. The hard work is authoritative offers, policy, attribution, identity and handoff - not a demo.

5. Plan Storefront Next after peak

   The software entitlement is included in B2C Commerce, but the migration is not free. Ask Salesforce for access to the announced AI Migration Toolkit and its installation guidance. Use it, if provisioned, to estimate effort before making the architecture decision.

The ecommerce leadership lens

• What are the numbers telling us? Baseline search exit, zero-result rate, product-list-to-detail click-through, add-to-cart, conversion, revenue per visitor and assisted conversion before changing anything.
• Where are we off plan? Segment by device, query intent, category, inventory state and traffic source. AI averages can hide bad journeys.
• What needs to be fixed first? Product truth, offer truth, measurement and fallbacks. A smarter interface amplifies bad catalog data faster.
• Who owns it? One business owner for conversion, one technical owner for reliability, named owners for product data, legal/privacy, service and channel operations.

02 / THE RETAILER ACTION PLAN

What to do before peak, through holiday and immediately after

Dreamforce produced a long list of announcements. This is the sequence I would use as a retailer-side ecommerce leader to turn that roadmap into practical action without putting peak revenue at unnecessary risk.

03 / DREAMFORCE ANNOUNCEMENT MAP

What Salesforce announced and the retail priorities

Dreamforce covered discovery, shopper agents, external AI channels, storefront modernization, operations, OMS, POS and B2B. This is the shortest useful version for an SFRA or SiteGenesis retailer.

Priority key

• P1: Start immediately, but use bounded scope, holdouts and rollback.
• P2: Valuable enablement or planning work with limited peak risk.
• P3: Only if it is already available, staffed and behind a feature flag.
• P4: Put it on the post-holiday architecture roadmap.

04 / FROM ANNOUNCEMENTS TO ADOPTION

Turning announcements into a funded roadmap

Dreamforce made the platform direction clearer. The harder question for an ecommerce leader is how to turn that direction into a funded roadmap, a production capability and a measurable business result.

What those ideas mean in commerce

• Trapped value: Product, customer, inventory, service and behavioral data only create value when they help a shopper or commerce team make a better decision. Connectivity is necessary; trusted context and a usable workflow are what make the data productive.
• Less human work: The opportunity is not to remove judgment. It is to remove repetitive research, rule maintenance, comparison and coordination so people can focus on exceptions, strategy and customer recovery.
• A changing interface: Search boxes, category pages and dashboards will not disappear overnight. But more journeys may begin with a stated need, with AI assembling the products, evidence and actions the shopper or merchant needs next.

The translation retailers need

Dreamforce makes the platform direction visible. Retail leaders still need a decision-ready view: the operating problem, a realistic implementation path, implementation requirements, ownership, risk and the outcome that should move. This guide provides that translation and makes clear what belongs before holiday, what needs proof and what should wait.

Sources: Salesforce AIforce announcement for the product framing; TechRadar keynote report for the Benioff quotation.

Translate capability into a decision an ecommerce leader can own

A more effective adoption program would begin with the business problem, prove the capability in a realistic commerce journey, and close the loop between customer evidence, investment and product direction.

A practical example: owning the product card

In a point-in-time test of the Crocs and HEYDUDE experiences after Salesforce's commerce announcement, I began with an unbranded shopping need. ChatGPT built the consideration set, understood the products and variants, and compared offers. Crocs.com won the click when the direct offer was competitive. For HEYDUDE, SCHEELS had the stronger offer and won the click.

The program I would build

1. Start with retail business problems

   Use search failure, merchandising effort, service demand, conversion friction or channel leakage - not a product name - as the entry point.

2. Show the capability in a realistic commerce scenario

   Use representative catalog complexity, variants, localized offers, inventory states, checkout and failure handling.

3. Make adoption visible

   Publish prerequisites, supported architectures, implementation guidance, timeline, KPI and named ownership.

4. Create the customer feedback loop

   Capture what retail teams value, what blocks production and what would justify the next investment; route that evidence into roadmap and enablement decisions.

05 / READ THIS CORRECTLY

How to read product status

I separate Salesforce product status from the work required to make a capability safe and productive in a real retail environment.

06 / PRODUCT DISCOVERY

Agentic Commerce Search is the clearest near-term bet

This is the feature I would put through a retailer-led pilot first because it addresses a measurable problem and offers a controlled path to proof.

What it means

• Moves discovery beyond literal keyword matching toward intent and conversational context across search, browse, recommendations and merchandising rules.
• Salesforce says B2C Commerce customers can activate it against the existing catalog; headless APIs extend it to other front ends.
• Salesforce documents it as a B2C Commerce toggle and also offers headless APIs. I found no published SFRA- or SiteGenesis-specific implementation pattern, so confirm the integration path in your realm and codebase.
• Shop the Store is an Agentic Commerce Search feature, GA September 2026, that lets shoppers browse products available at a selected physical store.
• Guided Discovery combines Agentic Commerce Search with Shopper Agent and requires both products. Validate the technical prerequisites for your implementation.
• In my assessment, it does not replace checkout, CMS/content operations, order management, CRM/service, analytics, external channel feeds or specialist capabilities such as retail media and cross-property content search.

Customer experience test

The interface is not the product. The product is the ability to preserve page context, catalog truth and the shopper's place in the journey while the agent helps compare or narrow options.

Customer experience test: Can the shopper recover gracefully when the agent is wrong, unavailable or slow?

07 / EXISTING SEARCH INVESTMENTS

Do not replace a good search platform just because Salesforce has a new one

Many SFRA and SiteGenesis retailers replaced the original out-of-box search for a reason. Agentic Commerce Search deserves a fresh evaluation, but it does not erase the value, tuning or switching cost already embedded in a mature third-party platform.

What native integration can improve

• Fewer vendors and potentially fewer custom synchronization, authentication and release dependencies.
• Closer alignment to the B2C Commerce catalog and Salesforce's Carter, merchandising and agentic-channel roadmap.
• A simpler escalation path when search, catalog and storefront behavior cross product boundaries.
• Potentially lower total cost if the third-party platform is used narrowly and Salesforce can replace the full production scope.

What it does not eliminate

• Catalog indexing, event collection, relevance tuning, monitoring, fallbacks and quality assurance still exist.
• Guided Discovery requires both Agentic Commerce Search and Shopper Agent. Native integration still needs a validated architecture and a supported implementation.
• Moving more of the journey to one platform reduces vendor handoffs but increases concentration risk when that platform has an incident.
• Affected, not replaced: a search-platform change can alter analytics instrumentation, retail-media workflows, SEO, personalization and merchant processes. In my assessment, Agentic Commerce Search remains a discovery layer; it does not replace checkout, CMS/content operations, OMS, CRM/service, analytics, external channel feeds, retail media or cross-property content search.

Five common enterprise replacements

This is not a vendor ranking. It is a practical recommendation for an SFRA or SiteGenesis retailer with a functioning production implementation.

Evaluate the complete customer experience

The right test is whether Salesforce can improve the customer's discovery experience while reducing the total operating burden at an acceptable migration risk.

The bakeoff I would require

1. Define equivalent scope

   Compare search, autosuggest, facets, browse, recommendations and personalization. Use the gap list on page 8 for the capabilities Agentic Commerce Search does not replace.

2. Run the same evidence

   Use the same catalog and head, long-tail, misspelled, natural-language and zero-result queries in a controlled traffic split or holdout.

3. Score customer and operating outcomes

   Measure conversion, revenue, margin and relevance alongside freshness, latency, availability, fallback, merchant effort and recovery time.

4. Model cost and require non-inferiority

   Assess implementation effort, ongoing maintenance and team capacity. Protect conversion, margin, relevance and reliability before counting consolidation benefits.

08 / ASSISTED SHOPPING

Carter should earn the right to expand

Carter is Salesforce's Agentforce shopper agent and one of seven job-ready agents announced on September 11. Salesforce describes it as supporting in-chat checkout. That can reduce friction, but the dangerous implementation is a general-purpose box with no journey ownership, no escalation and no measurement.

Availability snapshot

How I would implement it

1. Choose one expensive customer question

   Compatibility, room fit, product comparison, regimen building or replacement-parts discovery are better starting points than 'ask me anything.'

2. Define authorization

   Write down what the agent may recommend, what it may change, what needs consent and when it must hand off.

3. Protect commerce state

   Variant, price, promotion, inventory, fulfillment promise and cart state must survive the conversation and handoff.

4. Instrument the whole journey

   Measure assisted-to-PDP, assisted add-to-cart, conversion, containment, escalation, abandonment, error rate and incremental revenue - not chat volume.

5. Launch as a cohort

   Use limited traffic, selected categories and a kill switch. Expand only after quality and economics are visible.

09 / AGENTIC COMMERCE CHANNELS

Discovery is moving off your site; accountability is not

ChatGPT, Google AI Mode and Gemini are becoming important discovery surfaces. Transaction capability differs by channel, and the merchant still owns offer truth, customer trust, fulfillment and the order.

What was announced

• The Salesforce/OpenAI commerce partnership dates to October 2025, when Agentic Commerce Protocol and Instant Checkout were described as a future integration. Salesforce's June 24, 2026 release confirmed July GA for the Commerce catalog integration.
• The July retailer scope is a catalog feed with Business Manager field mapping and daily syncs. Confirm the current checkout path for your program: whether the experience hands off to your storefront or has access to a separately available Instant Checkout/ACP transaction path.
• The Google catalog feed integration is GA now. Retailers can map product fields in Business Manager and send the catalog to Google Merchant Center before peak.
• Google UCP integration is scheduled for October 2026 and is the capability intended to support checkout directly on Google surfaces while the retailer remains merchant of record.
• TikTok Shop integration is scheduled for January 2027. Shop the Store is not an external channel; it is a September Agentic Commerce Search capability for store-scoped browsing.

What has to be true before launch

• Stable product identifiers and complete variant relationships across feed, storefront, cart and order.
• Fresh price and availability with explicit latency tolerance and fallback behavior.
• Promotion and fulfillment logic that does not create a different promise in AI than at checkout.
• Attribution that distinguishes AI referral, assisted discovery, cart creation and completed order.
• Terms for returns, cancellations, substitutions, customer service, privacy, consent and disputed agent actions.
• Operational monitoring for feed freshness, rejected items, broken handoffs, duplicate carts and channel-specific error rates.

10 / MODERNIZATION

Storefront Next is an option, not a holiday deadline

Salesforce describes Storefront Next as generally available and included in B2C Commerce. Implementation still requires migration planning, testing and operational change.

Available now

• Storefront Next core entitlement. Salesforce also labels the AI Migration Toolkit as available now, but I could not find a public download or installation page.
• Salesforce says the toolkit supports route-by-route and feature-by-feature refactoring, letting teams validate new Storefront Next routes against the live legacy site. Ask your account team or support contact for access to afnext-migrate, its documentation and applicable terms.
• Hybrid Auth supports phased coexistence, so SFRA or SiteGenesis routes can continue running alongside Storefront Next during migration.
• Shipping and delivery estimates are GA, but they run on Order Management's Commerce Delivery Service. Validate that dependency and its integration with your storefront.
• Native Commerce Apps for Storefront Next, according to the Dreamforce innovation page.

Coming in the immediate release window

• New industry templates - September 2026.
• Order on behalf and Streamlined Guest Authorization - October 2026.

Request access, then use the toolkit as a planning instrument.

Inventory custom cartridges, templates, integrations, tracking, SEO dependencies, checkout changes, accessibility issues and operations tooling. If Salesforce provisions the toolkit, use its route-by-route analysis with Hybrid Auth to reduce big-bang risk while the legacy storefront continues operating. If access is not available, use the same inventory and route-based assessment manually.

My default: assess now; design and build after peak unless the new stack is already production-ready.

What the 'included' statement does and does not mean

11 / OPERATIONS AND PLATFORM

The quieter features may pay back faster

Some Dreamforce items are not flashy, but they can improve daily merchandising, release quality and incident response without changing the storefront architecture.

What I would take into peak

• Observability that reduces mean time to detect and recover.
• Cache and operational controls that have passed load and rollback testing.
• Merchant productivity improvements that do not alter the customer journey.
• Profile sync only after identity, consent and failure behavior are explicit.

12 / THE REST OF THE COMMERCE PORTFOLIO

Useful announcements, but not free B2C features

Dreamforce grouped B2B Commerce, Order Management and POS under the commerce story. Their innovations require those underlying products and should not be confused with your B2C storefront entitlement.

What the enterprise AI announcements mean for commerce

• Carter is the commerce-specific item to track. Salesforce included the shopper agent in its seven job-ready agents. For an ecommerce roadmap, Carter's availability, storefront integration, dependencies and measurable customer journey matter more than the larger portfolio label.
• AIforce is an architecture signal, not a holiday feature. Salesforce positions it as a way to expose platform capabilities through AI-driven interfaces. Commerce teams should use that direction to plan how product, customer and order capabilities can be accessed safely beyond fixed screens.
• The Trusted Enterprise AI Harness belongs in 2027 planning. Salesforce says its new capabilities and unified experience begin rolling out in early fiscal FY28, roughly February 2027. It may shape model choice, evaluation, auditability and governance, but it should not change the 2026 peak backlog.
• Confirm product names and prerequisites before implementation. Salesforce product names and capabilities are changing quickly. Use current documentation and validate availability in your environment.

What not to do

• Do not buy an AI platform narrative without a specific retail journey and KPI.
• Do not assume a cross-cloud demo equals a bundled entitlement.
• Do not put January 2027 roadmap items into a 2026 holiday forecast.

13 / HOLIDAY 2026

The operating sequence into peak

At this point in the calendar, scope discipline beats ambition. The objective is to enter peak with better discovery and stronger control, not a larger blast radius.

Recommended focus allocation

14 / MEASUREMENT AND OWNERSHIP

Make every feature answer to a retail outcome

AI activity is not an outcome. The operating dashboard should connect experience quality to conversion, margin, service and reliability.

Minimum ownership model

15 / IMPLEMENTATION READINESS

Prepare the team before rollout

A successful rollout depends on a clear customer use case, a supported implementation and a team ready to run it. Align with Salesforce and your implementation partners on the practical path from announcement to production.

1. Start with the customer problem

   Choose the journey that needs improvement and the business outcome you will measure.

2. Confirm the implementation path

   Validate availability, architecture fit and technical prerequisites in your own environment.

3. Make ownership clear

   Name the people responsible for product data, integration, measurement, service and ongoing support.

4. Prove readiness before expanding

   Test the complete journey, train the team and establish monitoring and a reliable fallback.

16 / SOURCES AND NOTES

Official product references

Availability and product positioning reflect public materials reviewed on 20 September 2026. Roadmap dates can change. Confirm current availability and technical prerequisites before implementation.