An Agentforce pilot is the only reliable way to produce the conversation-rate and action-profile baselines that the production commitment relies on. The Admodum read on the pilot scope, the success criteria, the measurement discipline and the graduation gate.
An Agentforce pilot is the baseline-production exercise. The pilot produces the two measured baselines that the production commitment relies on: the conversation-rate-per-audience-member and the action-profile-per-conversation. Without these baselines, the production commitment is sized against the Salesforce account-team's marketing-led estimate.
The pilot is the buyer-side discipline. The Salesforce account team will offer a pilot-program credit or a discounted pilot commitment; the credit or the discount is the commercial vehicle, not the methodology. The methodology is the buyer-side measurement plan. The wider Agentforce per-conversation arithmetic spoke reads the production commercial model that the pilot informs.
The pilot scope is one or two clearly-bounded use cases. The candidate use cases are typically a customer-service triage agent (a first-line agent that resolves a defined set of high-volume cases before handing off to a human agent), a sales-development outbound agent (an agent that runs early-stage outreach against a defined contact list) or a partner-network self-service agent (an agent that serves a partner audience with self-service answers).
The pilot is sized for measurement. A pilot scope that covers the platform-wide deployment in the first eight weeks does not produce a stable baseline; the audience cohort is too heterogeneous, the action profile is too varied and the measurement is too noisy. A two-use-case pilot produces two readable baselines.
The pilot duration is typically eight to twelve weeks. The minimum duration is the time required to produce a stable consumption rate against a representative audience cohort: the conversation rate stabilises only after the audience has had time to discover the agent and to use it for the second and third times.
The maximum useful duration is twelve to sixteen weeks. Beyond that, the pilot consumes commercial budget that anchors the negotiation around the pilot's small scale rather than the production deployment's larger scale. The pilot graduates at the point at which the baselines are readable.
The success criteria gate the graduation from pilot to production commitment. The first criterion is the conversation-rate baseline against a representative audience: a stable number of conversations per audience member per week or month. The second criterion is the action-profile measurement: the typical action count per conversation, broken down by record-read, record-write, flow execution and Data Cloud query.
The third criterion is the end-user outcome metric: the containment rate for a customer-service triage agent, the conversion-rate uplift for a sales-development agent, the time-to-resolution for a service agent. The fourth criterion is the integrated-cost baseline: the conversation cost plus the action-credit cost against the outcome metric, producing the cost-per-outcome that the production commitment is sized against.
The measurement discipline reads the conversation log, the action log and the outcome data in parallel. The conversation log produces the conversation-rate baseline; the action log produces the action-profile baseline; the outcome data produces the end-user outcome metric. The three are joined to produce the per-conversation integrated cost-per-outcome.
The buyer-side artefact is the pilot-evidence pack: the conversation-rate per audience member per week, the action profile per conversation, the outcome metric, the integrated cost-per-outcome and the variance against the Salesforce account-team's pre-pilot forecast. The wider Data Cloud sizing spoke reads the parallel discipline on the data-platform side.
The graduation gate is the point at which the pilot baselines feed the production commitment design. The graduation produces three artefacts: the conversation-rate baseline, the action-profile baseline, the per-conversation integrated cost. These three feed the moderate-case commitment forecast that the wider Agentforce per-conversation arithmetic spoke describes.
A pilot that does not graduate (the conversation rate is below the threshold, the action profile is uneconomic against the outcome metric, the integrated cost-per-outcome is above the human-served benchmark) is a buyer-side success: the production commitment is not signed against the wrong use case. The audit-defence-grade discipline reads the pilot evidence pack against the no-go decision with the same rigour as it reads the production-readiness decision.
The production commercial model that the pilot baselines feed.
The parallel discipline on the data-platform side that the agent draws on.
The predecessor AI-feature surface and its interaction with the Agentforce stack.
A senior Admodum Salesforce advisor will read your pilot scope, your measurement plan and your graduation criteria on a private call. Active deployment moments route to Renewal Programme.