Which Agentforce Voice Meter Should We Put on the Order Form: Voice Actions or Voice Minutes?

Salesforce meters Agentforce Voice two ways. Voice Actions sit on the main Flex Credits Rate Card. Voice Minutes sit on a supplemental rate card. They are mutually exclusive on the Order Form: if you purchase Voice Minutes, the Voice Actions usage types on the main rate card do not apply. Sources: Flex Credits Rate Card PDF and Voice Minutes Supplemental Rate Card PDF.

This is not Flex Credits versus Conversations. It is not Help Agent resolution metering. It is not PayGo versus Pre-Commit versus Pre-Purchase. It is which Voice usage type your AE writes into Usage Details once you already chose Flex Credits and you are putting phone agents live. Sibling forks: Flex vs Conversations, Help Agent 400 credits, PayGo vs Pre-Commit vs Pre-Purchase.

Competitor explainers already rank on this fork. Cyntexa and Ashapura Softech both walk Voice Actions versus Voice Minutes and the break-even math while our live /llms.txt Questions set covers Flex Credits buying models, Help Agent, AIforce, and Wallet without a Voice meter decoder. Demand sources: Cyntexa and Ashapura Softech.

The two meters (vendor numbers)

List multipliers only. Contracted rates come from your AE. Flex Credits list is $500 per 100,000 credits on the Agentforce pricing page, which is $0.005 per credit for planning math. Source: salesforce.com/agentforce/pricing.

1. Voice Actions (main rate card, updated as of August 31, 2026). Standard Voice Action and Custom Voice Action: 30 Flex Credits in production, 24 in sandbox. Footnote 1 on that card: for certain customers, the Agentforce Voice Minutes usage type may apply instead. Source: Flex Credits Rate Card PDF.
2. Voice Minutes (supplemental rate card, updated as of February 3, 2026). Agentforce Voice Minutes: 60 Flex Credits per minute in production, 48 in sandbox. Explicit exclusivity line: The Agentforce Voice Actions usage types set forth on the Flex Credits Rate Card do not apply to Customers that have purchased Agentforce Voice Minutes and instead the Agentforce Voice Minutes usage type applies. Source: Voice Minutes Supplemental Rate Card PDF.

At list: one Voice Action is about $0.15. One production Voice Minute is about $0.30. Sandbox is cheaper on both cards (24 and 48). Same PDFs.

Salesforce's pricing FAQ points Voice pricing back to the rate card rather than publishing a third number: Usage types and multipliers differ for Agentforce Voice actions. See the Agentforce Rate Card for more detail. Source: salesforce.com/agentforce/pricing.

Mutual exclusivity is the Order Form decision

You do not run both meters on the same entitlement. The supplemental card is the hard line: purchase Voice Minutes and Voice Actions stop applying. The main card's footnote points certain customers to Voice Minutes. Put the chosen usage type on the Usage Details table before signature, not after the first Digital Wallet surprise. Sources: both rate cards above.

Telephony sits outside this fork. Service Cloud Voice, Amazon Connect, or bring-your-own telephony are separate line items. Model AI credits and telephony on two rows, not one. Primary sources above do not price telephony; treat partner explainers as demand evidence, not as Salesforce list.

Break-even worksheet (list math)

Run this on a sample of real call transcripts before the AE call. Use Digital Wallet or a pilot tag once live; do not invent action counts from a demo.

1. Pull 50 to 100 recent voice interactions. For each call, count discrete Voice Actions the agent would fire (identify caller, look up record, answer, update, finalize).
2. Record talk time in minutes. Confirm with your AE whether Voice Minutes round up to the next full minute on your Order Form. The supplemental card meters Per minute; rounding rules belong in writing.
3. Credits under Voice Actions = actions × 30 (production).
4. Credits under Voice Minutes = minutes × 60 (production).
5. Prefer Voice Actions when actions × 30 is lower than minutes × 60 across the p50 and p90 call. Prefer Voice Minutes when action density is high enough that minutes × 60 wins.
6. Worked list example (not a recommendation): a 3-minute call with 4 Voice Actions costs 120 credits under Actions ($0.60 at list) versus 180 credits under Minutes ($0.90 at list). Same 3-minute call with 8 Voice Actions costs 240 credits under Actions ($1.20) versus 180 under Minutes ($0.90). The fork flips with action density, not with feature names.

Sandbox multipliers (24 / 48) apply in pre-production, including scratch orgs per the main card. Budget build and test separately from production. Sibling: Sandbox Flex Credits rate.

When each meter wins for a Director+ team

1. Short, low-action calls (identity, status, one update): Voice Actions usually win on list math because minutes still burn even when the agent fires few tools.
2. Long, high-action calls (multi-step troubleshooting, several lookups and updates per minute): Voice Minutes can win once actions per minute push Actions credits above Minutes credits.
3. Unknown mix: start on Voice Actions only if the AE confirms that is the default on your Order Form and Digital Wallet will show Voice Action burn. Do not assume Minutes until Usage Details names Voice Minutes.
4. Always separate telephony. AI Flex Credits and phone platform fees are different owners. Name RevOps for the credit model and whoever owns Service Cloud Voice / CCaaS for the phone line.

Unused Flex Credits still do not roll over. That rule is on the Agentforce pricing FAQ and both rate cards. It does not change which Voice meter you pick; it changes how hard you size the pack once you pick. Sibling buying-model worksheet: PayGo vs Pre-Commit vs Pre-Purchase.

What to put in writing before you sign

1. Chosen Voice usage type: Voice Actions or Voice Minutes, copied from Usage Details.
2. Production and sandbox multipliers as they appear on the attached rate card version (date stamp the PDF).
3. Confirmation that the other Voice usage type does not apply under that purchase.
4. Whether minutes round up, and who owns Digital Wallet alerts for Voice burn.
5. Sample worksheet: p50 / p90 actions per call, p50 / p90 minutes, list and contracted credit cost under both meters.
6. Kill criteria if Voice burn diverges from the model: metric, floor, date, consequence. Sibling: AI SDR kill criteria.

What the sibling decoders already cover (do not duplicate)

• Flex Credits versus Conversations break-even: Breakeven
• PayGo / Pre-Commit / Pre-Purchase: PayGo
• Help Agent 400-credit resolution: Help Agent
• Sandbox multiplier: Sandbox
• Digital Wallet has no hard kill switch: Digital Wallet
• Tokens in the AI Usage Data Model versus Flex Credit bills: Usage Data Model

This page is the Voice meter on the Order Form once those forks are decided.

Related: Breakeven · PayGo · Sandbox · Digital Wallet