Salesforce Agentforce is priced per conversation rather than per user. The Admodum read on the conversation metric, the consumption forecast, the action-credit interaction and the buyer-side discipline that sizes the commitment to the actual deployed agent population.
Agentforce is priced per conversation. A conversation is the multi-turn interaction between an end-user (a customer, an employee, a partner) and a deployed Agentforce agent. The metering unit is the conversation, not the message: a five-message exchange and a fifty-message exchange both count as one conversation.
The metering is the directional shift away from the per-user metric that governs Sales Cloud, Service Cloud and Marketing Cloud. The shift recognises that an agent serves a population (the customer audience, the employee population, the partner network) rather than a defined seat list. The wider Sales Cloud editions spoke reads the parallel seat metric.
The launched price point is in the order of two dollars per conversation. The realised price is banded: a higher-volume commitment buys a lower per-conversation rate; a lower-volume commitment buys a higher per-conversation rate. The bands are negotiated against the conversation commitment volume, the bundled action-credit allowance and the wider commercial position.
The contractual artefact reads three numbers: the per-conversation rate at the contracted band, the conversation commitment for the contract year, the overage rate that applies to conversations above the commitment. The wider Data Cloud consumption credits spoke reads the parallel consumption-credit mechanics for the data-platform layer.
Each conversation triggers one or more actions. An action is a record-read, a record-write, a flow execution, a Data Cloud query, an external API call. The action consumption is metered against an action-credit allowance that ships with the conversation commitment.
The action-credit allowance is the silent secondary metric. A conversation-light, action-heavy agent (a customer-service agent that resolves a complex case via multiple record-writes and flow executions) consumes the action-credit allowance well above the conversation-count proportion. The buyer-side discipline reads the action profile of each deployed agent against the action-credit allowance.
The deployed-agent inventory is the sizing baseline. The inventory reads each deployed agent (the customer-service triage agent, the sales-development agent, the field-operations support agent, the partner-network self-service agent) and the conversation forecast against the audience that the agent serves.
The forecast reads three inputs per agent: the audience size (the customer count, the employee count, the partner count), the conversation-per-audience-member rate (a quarterly conversation rate for a customer-service triage agent, a weekly rate for a sales-development agent), the action profile (the typical action count per conversation). The wider Agentforce pilot spoke reads the pilot methodology that produces the conversation-rate baseline.
The commitment-design discipline is the workload-by-workload forecast. The forecast reads each deployed agent, the conversation rate per agent, the action profile per conversation, the per-conversation rate at the contracted band and the overage rate. The output is the moderate-case commitment that aligns to the actual deployed agent population.
The buyer-side discipline reads the overage rate as the principal contract artefact. An overage rate at or near the realised per-conversation rate is a tolerable position: the buyer can absorb above-commitment conversations at the same effective rate. An overage rate at a multiple of the realised rate is a renewal-cycle risk: the buyer is exposed to a price step at the moment that adoption exceeds the commitment.
The renewal posture is driven by the actual-consumption audit. The audit reads the conversation rate against the commitment, the action-credit drawdown against the allowance and the agent-by-agent contribution. The renewal-cycle commitment is sized against the actual consumption baseline rather than against the prior year's optimistic forecast.
The renewal evidence pack reads four artefacts: the conversation-rate run-rate, the action-credit drawdown rate, the agent-by-agent deployment inventory, the moderate-case forecast for the new contract year. The discipline aligns the commitment to the deployed agent population rather than to the marketing-led roadmap forecast. The wider seat-reassignment policy spoke reads the parallel discipline on the seat-based contracts.
The methodology that produces the conversation-rate and action-profile baselines.
The parallel consumption-credit model for the data-platform layer.
The predecessor AI-feature licensing surface that runs alongside Agentforce.
A senior Admodum Salesforce advisor will read your Agentforce conversation forecast, your action-credit consumption and your commitment-design artefact on a private call. Active renewal moments route to Renewal Programme.