Should Salesforce kill the per-seat pricing model?
Recommended path: Hybrid transition. Kill pure per-seat for new logos by 2028; preserve legacy per-seat at premium (enterprise floor). Migrate core CRM to consumption-based units ("Salesforce Credits" anchored to API calls + data volume + feature tier), bolt Agentforce on consumption.
Specific moves: (1) Announce "Flexible Consumption Era" at Dreamforce 2026; (2) pilot consumption-pricing tier at 50+ enterprise accounts Q4 2026; (3) grandfather existing per-seat deals at locked rates through 2030; (4) price agent-driven workflows at $3-8K/month per agent (consumption-based, not seat-based).
Why Per-Seat Is Broken
- AI substitution deletes seats mechanically. Agentforce + Einstein Copilot automate 30-40% of entry-level CRM tasks (data entry, lead scoring, opportunity qualification). Each AI deployment = RIF wave = fewer licensed users per customer. Per-seat revenue collapses as headcount falls. ServiceNow and Snowflake experienced this exact erosion 2023-2025.
- Customers hide true headcount. Seat hoarding creates friction: managers fear licensing more users for fear of cost. Customers deploy via shared logins, temp accounts, read-only access—shadow licensing. Salesforce can't invoice what it can't measure. Consumption eliminates the perverse incentive.
- Competitive pricing arbitrage kills margins. HubSpot Sales Hub $50-120/user/mo, Pipedrive $15-99/user/mo. Salesforce per-seat ($155-300+ depending on edition) undercuts smaller SMBs into cheaper alternatives. Consumption-based locks in enterprises ("you use what you use"), avoids per-user comparison shock.
- AI agents create pricing paradox. A single Agentforce agent replaces 2-3 human reps but licenses as $0 under per-seat (it's not a "seat"). Salesforce can't monetize the productivity shift. Consumption-based pricing (e.g., per-agent interactions, per-API-call) fixes this.
- Margin compression from platform consolidation. Enterprises buying Tableau + Data Cloud + Slack alongside core CRM negotiate bundled per-seat discounts across entire stack. Per-seat model encourages mega-discounts. Consumption (separate metering per product) restores granular pricing power.
- Talent cost inversion flips the narrative. By 2027, average Salesforce user cost (loaded rep salary + onboarding + admin overhead) will exceed $200K/year. Customers increasingly ask "why pay per seat for a tool that now handles 50% of that work?" Consumption-based pricing shifts answer: pay only for actual CRM transactions, not headcount.
Why Salesforce Won't Kill It Outright
- Enterprise ACV lock-in. ~60% of Salesforce's $0-15M ARR deals are per-seat; these renew predictably. Cold-turkey migration to consumption creates renewal chaos, churn spike, and analyst downgrades. Marc Benioff's job depends on guidance. Hybrid (grandfather legacy, transition new logos) spreads risk.
- Large-scale billing system lift. Salesforce's billing engine (Zuora-backed) runs on per-seat assumptions across invoice generation, usage reporting, compliance (SOX, GDPR audit trail). Full consumption pivot requires 18-24mo rewrite and parallel-run validation. Too expensive/risky in one go.
- Sales org friction. Salesforce's 10K+ quota-carrying reps have 30-year mental model of "per-user = quota." Sales rep earning plans, commission structures, and customer ROI calculators all hinge on seat counts. Switching to consumption mid-career breaks rep productivity and comp predictability.
What Salesforce Should Actually Do
- Launch "Salesforce Credits" consumption tier Q4 2026. Announce at Dreamforce: "Starting 2027, new Salesforce Customer accounts can choose Salesforce Credits (consumption-based) or traditional per-seat licensing." Position credits as "modern pricing for modern AI-driven workflows." Pilot at 50-100 enterprise Fortune 500 accounts.
- Price Agentforce exclusively on consumption. Agentforce agents = $2,500-8,000/month per agent (not per seat). Agents handle up to 100 concurrent conversations. Salesforce monetizes the productivity, not the headcount.
- Metric: API calls + data volume + feature consumption. 1 million API calls/month = base tier ($10K/mo); +storage overage ($0.50/GB/mo); +Einstein Einstein Copilot interactions (+$5K/mo). Transparent, usage-fair, removes hidden billing shock.
- Grandfather existing per-seat deals through 2030. Lock current contract rates; auto-renew at per-seat pricing for customers who want it. Zero forced migrations. This removes churn risk and gives sales org 4-5 years to retrain.
- Separate Agentforce P&L from core CRM. Run Agentforce as distinct Business Unit with consumption-based unit economics. Prove the model works before rolling into core. Gives board + analysts clearer visibility into "legacy vs. Growth" revenue split.
- Invest in consumption-based ROI calculators. Partner with Pavilion, Bridge Group, Klue, and Force Management to build ROI case studies showing "Customer X paid $3.2M/year per-seat; migrated to Agentforce + Credits; now pays $1.8M/year for 3x the capacity." Sell the narrative, not the fear.
- Negotiate reseller and partner pricing tiers. AppExchange partners need clear SaaS margin protection: e.g., Salesforce keeps 30%, partner keeps 70% on incremental consumption. Prevents channel conflict; grows partner adoption of Agentforce.
- **Announce 2027 sunset date for *cheapest* per-seat tier.** Kill "Salesforce Platform" ($165/user/mo) by 2027; keep "Sales Cloud" and "Service Cloud" per-seat at premium ($250+/user/mo) through 2030. Forces low-touch/mid-market customers toward consumption; protects high-margin enterprise per-seat.
| Pricing Model | Today (2026) | 2027 Roadmap | 2030+ Vision | Customer Reaction | Margin Impact |
|---|---|---|---|---|---|
| Per-Seat (Sales/Service Cloud) | $155-300/user/mo; ~$15-20B ARR | Maintained for enterprise; $250+/user/mo floor | Gradually phased out; legacy only | Enterprise keeps; resents caps on AI scaling | Stable; slight compression from hybrid migration |
| Consumption (Credits/Agentforce) | Pilot; <1% revenue | 15-25% of new customer ARR | 40-60% of total ARR | SMB + new logos adopt quickly; enjoy flexibility | High-margin; $0.08-0.20 per unit delivered |
| API + Data Volume | Not separately metered | Priced transparently ($10K base + $0.50/GB) | Core model | IT buyers love granularity; CFOs budget more predictably | +12-18% gross margin vs. per-seat discounting |
| Agentforce Agents | <0.5% adoption | $3-8K per agent/month | $5-12K per agent/month (competitive with 2x human reps) | CROs excited; procurement nervous about "another line item" | 65-72% gross margin; 40% incremental EBITDA |
| Slack + Tableau + Data Cloud | Per-seat discounts | Unbundled from core CRM licensing | Separate P&L; consumption-based optional | Buyers prefer modularity; less sticker shock | Margin recovery; +8-10% blended gross margin |
| Professional Services | Fixed + T&M | Bundled consumption packages | Outcome-based ("pay for agent productivity") | Customers prefer consumption-linked PS cost | 50-55% service margin |
Mermaid: Pricing Transition Path
FAQ
What is the "Salesforce Credits" consumption tier the article proposes? Salesforce Credits is a consumption-based pricing unit anchored to API calls, data volume, and feature tier, recommended for launch at Dreamforce 2026 with a pilot at 50-100 enterprise Fortune 500 accounts in Q4 2026.
The example pricing sets 1 million API calls per month as a base tier at $10K per month, plus $0.50 per GB storage overage and $5K per month for Einstein Copilot interactions. It is positioned as modern pricing for AI-driven workflows.
Why does per-seat pricing break down as AI agents are deployed? Agentforce plus Einstein Copilot automate 30-40% of entry-level CRM tasks like data entry, lead scoring, and opportunity qualification, so each deployment becomes a RIF wave that lowers licensed users and collapses per-seat revenue.
A single Agentforce agent replaces 2-3 human reps but licenses at $0 under per-seat because it is not a seat. ServiceNow and Snowflake experienced this exact erosion between 2023 and 2025.
How would Agentforce agents be priced under the new model? Agentforce would be priced exclusively on consumption at $2,500-8,000 per month per agent rather than per seat, with each agent handling up to 100 concurrent conversations. This lets Salesforce monetize the productivity shift instead of headcount.
The article also recommends running Agentforce as a separate P&L and business unit to prove the consumption model before folding it into core CRM.
Why won't Salesforce kill per-seat pricing outright? About 60% of Salesforce's $0-15M ARR deals are per-seat and renew predictably, so a cold-turkey migration would cause renewal chaos, churn spikes, and analyst downgrades that threaten Marc Benioff's guidance. The Zuora-backed billing engine assumes per-seat across invoicing, usage reporting, and SOX/GDPR compliance, requiring an 18-24 month rewrite.
Plus, 10,000-plus quota-carrying reps have comp plans and ROI calculators built on seat counts.
How does Salesforce's per-seat pricing compare to competitors and what gets sunset? Salesforce per-seat runs $155-300-plus depending on edition, undercutting SMBs into cheaper options like HubSpot Sales Hub at $50-120 per user per month and Pipedrive at $15-99 per user per month.
The plan announces a 2027 sunset for the cheapest tier, killing "Salesforce Platform" at $165 per user per month while keeping Sales Cloud and Service Cloud per-seat at a $250-plus premium through 2030. This pushes low-touch and mid-market customers toward consumption while protecting high-margin enterprise per-seat.
Bottom Line
Per-seat pricing is structurally at risk from AI automation. Salesforce won't kill it cold-turkey (enterprise lock-in + billing system complexity + sales rep friction too high), but the company *must* pivot new revenue to consumption-based models by 2027 to protect gross margin as per-seat ARR declines.
Agentforce is the Trojan horse: position agents as consumption-only, prove the economics work, then migrate core CRM on a 3-5 year glide path. Hybrid model (grandfather legacy, incentivize new consumption) is the only path that avoids 15-20% churn and keeps analysts happy.