COMMERCE / WHO GAINS? / WHO DEPENDS ON WHOM?
What gives a commerce business bargaining power with an AI partner?
A commerce business has more negotiating room when the AI partner needs its contribution and the business can reach customers elsewhere. Keeping checkout or joining several channels helps only when it changes those alternatives.
Assess the alternatives ↓PROVIDER’S ALTERNATIVE CUSTOMER ROUTES
01 / SCARCE COMPLEMENTS
The change in bargaining dependence
An AI channel can bring a retailer orders without becoming dependent on that particular retailer. It still needs someone to accept and fulfil purchases, but it may be able to direct shoppers to another seller. The retailer has more negotiating room when shoppers want its particular products, account benefits or service, making replacement costly for the channel. The contribution must give shoppers a reason to choose that retailer over alternatives.[1]
02 / PRACTICAL OUTSIDE OPTIONS
How practical alternatives change the balance
Several partnerships reduce dependence only if they offer a usable alternative for the same purpose. Instacart's six recorded AI experiences illustrate the distinction: a route that hands shoppers to Instacart can offer another way to acquire customers, but it does not reproduce payment inside an AI channel. The table below separates those roles.[2]
WORKED EXAMPLE / INSTACART
A route to shoppers is not always a substitute for checkout
Six AI experiences provide documented routes to Instacart. Only the ChatGPT launch documents payment inside the channel; its later continuity is unresolved. The alternatives depend on what Instacart needs the relationship to do.
| Objective | What is documented | What to establish |
|---|---|---|
| Reach shoppers | Six experiences, with different dates and statuses. | Do they reach relevant buyers at usable scale? |
| Take payment inside AI | ChatGPT’s launch design. Current continuity unresolved. | Which routes can provide this capability now? |
| Connect an existing account | Account connections in four configurations. | What benefits, data and exit rights remain usable? |
Inspect the dated routes and evidence limits
| Objective | Documented routes | Limits | Unknown |
|---|---|---|---|
| Reach shoppers in a conversation | ChatGPT app (launched December 2025); Claude connector (announced April 2026); Gemini connected app (May 2026); AI Mode (July 2026); a Copilot recipe button and a Muse connector | Dates and statuses differ; launch evidence is not current availability | How many shoppers use each route and whether they would switch |
| Complete payment inside the channel | Only the ChatGPT launch design: payment inside Instacart’s app on OpenAI’s Instant Checkout component | ChatGPT operator and post-March payment continuity unresolved. Claude, Gemini and AI Mode hand off; Copilot and Muse checkout locations are undocumented. | Whether any channel currently completes the order in-channel |
| Keep the customer account connected | Instacart sign-in (ChatGPT); connected account (Claude, Gemini); account and membership benefits (AI Mode) | Account linking does not establish retention | Data-sharing and exit terms |
Claude, Gemini and AI Mode document a handoff to Instacart. Copilot and Muse checkout locations are not documented. Account benefits support repeat purchasing when they remain useful in the new channel. Alternative routes strengthen the provider's position when they reach the same buyers on workable terms.
Research basis: CP-6 Q5 and CP-4 configuration records, through 7 October 2026. These describe dated arrangements, not current availability.
Inspect the company’s dated configurations →Compare checkout roles and their supporting evidence →For an AI supplier, test the workflow instead. A second model is an alternative only if it can meet the same requirements. Compare models on the same task and operating requirements to assess whether they are substitutes.
03 / THE TERMS OF PARTICIPATION
Strategic implications for partnership terms
For a retailer, an alternative route is useful in a negotiation only if the business can use it without losing the customers, account benefits or service the agreement supplies.[3] Account linking can carry existing benefits into a purchase, giving shoppers a reason to return when those benefits remain valuable. Catalogue permissions, activation rules and exit terms determine what the business can offer and where it can move. These rights strengthen negotiating room when they give the business a credible alternative to the partner's terms.[4]
EXPLORE THE RESEARCH
The arrangements behind the analysis
Inspect the recorded configuration and its competitive interpretation. Several records can describe the same partnership at different stages or in different roles.
Browse the full partnership research
Each record identifies the operating status and date of the arrangement.
CONNECT THE ARGUMENT
Strategic assessment
The strategic exposure is greatest when a channel can replace the retailer more easily than the retailer can replace the channel's customers. Assess the actual loss on each side, including account benefits and usable alternatives. Then examine how standards and changing customer habits could alter that balance.
Next: what changes the advantage? →Other ways to explore Commerce
Research records
Based on the reconciled 81 episode/role records across 58 direct case bundles, including documented, planned, historical and specification material. Seven company-context cases and one boundary case remain outside the direct distribution. Source passages, dates and configuration history remain in the coordination-pattern record panels.
APPLY THE RESEARCH
Assess what these partnerships mean for your position.
Work through the capabilities, contributions and terms that could strengthen your business—or commission a focused investigation.
Sources and notes
Theory. Richard M. Emerson, "Power-Dependence Relations," American Sociological Review 27, no. 1 (1962), pp. 31-41, especially pp. 32-33. The comparison concerns the valued outcome each partner supplies and costly alternatives for that outcome. The commerce application is the Observatory's interpretation; potential leverage is distinct from observed negotiated terms.
Case evidence. Partnerships Observatory, comparison of six dated Instacart AI experiences, research cutoff 7 October 2026, reproduced in the worked example on this page; Instacart, "Instacart App Launches in OpenAI ChatGPT - First Company to Offer New Instant Checkout App Experience" (8 December 2025). The comparison separates customer acquisition, in-channel payment and account connection; dates and statuses differ. Each configuration retains its recorded date and operating status.
Third-party analysis. Boston Consulting Group, "Agentic Commerce Is Redefining Retail - How to Respond" (2025), "The Threats to Retailers" and retailer-response sections. Its analysis of customer access and loyalty informs the account-continuity question. Whether an alternative preserves the desired result is the Observatory's application of the dependence argument, informed by BCG's account of customer access.
Case evidence. Shopify Help Center, "Selling on Microsoft Copilot," archived 5 October 2026, checkout deactivation, catalogue access and merchant sign-in provisions; Target Corporation, conversational AI fact sheet (June 2026), account linking and Circle benefits. These sources describe participation and account rules. The Observatory's Commerce research has a 7 October 2026 cutoff.