Salesforce’s Outcome-Based Pricing for Agentforce Help Agent: What Enterprise Buyers Need to Negotiate Now

Salesforce made a significant announcement about its AI agentic offerings: a move to outcome-based licensing. While this certainly feels like Salesforce has been listening to their customer base and seems to signal an increased level of confidence in the results being received as pitched, there are still things that all enterprise customers need to be aware of before signing on the dotted line.

Enterprise customers still need to understand the mechanics, run an effective negotiation (preparation and execution), craft and deliver the right messaging internally and with Salesforce, and secure full transparency is in place.

All of that must happen while protecting, and ideally leveraging, the broader Salesforce commercial relationship already in place. That relationship typically spans other licensing metrics, additional products such as Sales, Marketing, Commerce, MuleSoft, and Tableau, and multiple associated order forms. It may also require navigating a SELA construct, which adds its own layer of complexity, negotiation prep, and execution effort.

How Agentforce Help Agent Pricing Works

Agentforce Help Agent, generally available this month, is an autonomous AI-driven service agent built to help enterprises resolve their customers’ issues. It is priced at $2 per successful resolution. That $2 figure is not new; it carries over directly from the earlier conversation-based model, where Agentforce pricing was $2 per conversation.

Resolutions can be purchased in prepaid packs, with a minimum purchase of 1,000 resolutions, and enterprise buyers should treat that $2 rate and those pack minimums as a starting point for negotiation rather than a fixed price.

Of course, this is Salesforce’s “standard” or list pricing. And for all enterprise customers, “standard” should not be sufficient.  Enterprise customers need to ensure they effectively negotiate the size and price of resolution packs, volume discounting structures, in-term price protections, and renewal price protections.

What Actually Counts as a Resolution for Agentforce Help Agent

Under Salesforce’s outcome-based licensing, Salesforce customers only pay when an issue is resolved from start to finish autonomously. Additionally, any actions that happen in a 10-minute window during a call and in a specific window for a chat count as a single resolution regardless of the number of questions.

Salesforce has indicated that a resolution does not occur if there is negative customer feedback, such as an indication that the customer did not get the answer they wanted, if the customer asks for human interaction, or if the case is escalated to the service desk. If any of those occur, there is no charge at all, which is the real value proposition behind pay-per-resolution: the customer does not pay when the issue is not resolved.

Enterprise buyers should still push Salesforce to further define and give examples of when a resolution occurs and when it does not, especially as it pertains to the enterprise customer’s business. This should lead to clear definitions in the order form itself, along with language covering dispute resolution, billing verification reporting, and recourse if the reported resolution rate does not match the customer’s experience. Vague or informal assurances should not be sufficient.

It is also worth pointing out that Data 360 and Agentforce usage tied to this offering are unmetered, which removes one layer of tracking complexity.  And for those that are currently Agentforce or Data 360 customers, they are very aware of this complexity and the cost implications of not structuring those deals appropriately.

Why the Flywheel Still Matters

Outcome-based does not mean consumption-free. There is still a flywheel driving usage and cost, even though the metric has shifted from conversations to resolutions outcomes. As adoption grows and the AI service agent is used and relied upon to solve customer issues, the volume of needed resolutions will increase as well as the associated costs.

Enterprise Customers who assume this model eliminates the need for commercial rigor are setting themselves up for cost surprises down the road. The negotiation priorities from consumption-based deals still apply for the most part: volume discount structures that scale as usage grows, renewal price protection that survives beyond the initial term, and firm commitments from Salesforce rather than informal assurances. None of that goes away just because Salesforce seemingly listened to their customers and moved to outcome-based pricing where a customer only pays when there is resolution.  Enterprise customers want resolutions to happen, and happen often. So does Salesforce.

What This Means for Your Salesforce Negotiation

A move to outcome-based pricing, could be a positive step for enterprise customers relative to the often-problematic consumption-based models that are prevalent in AI product licensing, but positive does not mean finished, and it certainly does not mean enterprise customers can treat this as a set-it-and-forget-it purchase.

Every Agentforce Help Agent deal sits inside a larger Salesforce relationship that includes other licensing metrics, multiple products, and often multiple order forms, and treating this deal as an isolated line item and negotiation is a mistake that will cost leverage later. Enterprise customers should evaluate their Salesforce commercial relationship and use this point in time and interest in a very important product to Salesforce to ensure the right pricing, protections, commitments…etc., are in place across the board.

UpperEdge helps enterprise customers negotiate Salesforce and other major software and AI agreements to secure the right transparency, pricing, protections, and flexibility, while improving the go-forward relationship. Contact UpperEdge to prepare for your next Salesforce negotiation.