How should ServiceNow price Now Assist against Microsoft Copilot in 2027?
ServiceNow cannot win a per-user-per-month price war against Microsoft Copilot in 2027 — Microsoft's M365 install base + Office bundling makes their effective marginal cost on Copilot 365 approach $0, while ServiceNow carries real infrastructure cost on every Now Assist execution.
The right play is to reframe Now Assist as an *agent platform*, not an *AI add-on*, and never ship a $30/user standalone SKU that invites direct Copilot comparison. Three pricing moves for FY27: (1) keep Now Assist bundled inside the Pro Plus uplift (~30% over Pro), (2) layer AI Agent Studio on top with consumption pricing (per-execution / per-token), (3) introduce workflow-specific outcome SKUs — "Now Assist for Customer Service" priced per-resolved-ticket, "Now Assist for HR" priced per-onboarded-employee.
The one anti-pattern: do *not* unbundle Now Assist into a standalone per-user SKU — Salesforce already proved this fails (Einstein 1 standalone). The talk track vs. Microsoft is workflow context: Copilot summarizes your inbox; Now Assist resolves your incident, provisions your laptop, and closes your CMDB ticket — and prices on the *outcome*, not the seat.
The Pricing Math Today (May 2026 list)
- Microsoft Copilot 365: ~$30/user/mo, requires M365 E3/E5 underneath (~$36-57/user/mo) — total stack ~$66-87/user/mo
- Microsoft Copilot for Sales: ~$50/user/mo (or $20 with Sales Premium)
- Microsoft Copilot for Service: ~$50/user/mo standalone
- Microsoft Copilot Studio: ~$200/tenant/mo + message-pack consumption (~$0.01/message)
- ServiceNow Now Assist: bundled in Pro Plus tier — ~30% uplift over Pro, which on a per-employee basis maps to roughly $80-200/employee/yr, or ~$7-17/user/mo equivalent when normalized
- ServiceNow AI Agent Studio: rolling out with consumption-style pricing on agent executions (token + run metering)
Why ServiceNow CAN'T Match Microsoft On Per-User Price
- Microsoft has ~400M paid M365 seats — Copilot is incremental ARPU on installed base; their *marginal* cost to add a Copilot seat is near-zero on the distribution side
- Copilot inference is subsidized by Microsoft's own Azure capacity — ServiceNow is a *renter* of compute (NVIDIA H100s, partner clouds), not a hyperscaler
- Microsoft sells Copilot through the existing E3/E5 EA — no new procurement cycle, no new vendor onboarding, no new SOC2 review
- IT-buyer scenarios where Copilot wins on price: knowledge worker productivity, email/Teams summarization, SharePoint search, generic doc generation — anywhere the *system of engagement* matters more than the *system of record*
- ServiceNow per-employee pricing (vs. Per-named-user) means racing Copilot to $30/user actually *raises* ServiceNow's effective bill at large enterprises (10k employees × $30 = $3.6M/yr, vs. Current Pro Plus uplift math of ~$1-2M)
- A standalone Now Assist SKU would also cannibalize the Pro Plus uplift that Bill McDermott has explicitly told the Street is the FY26-27 growth lever
The 3 Pricing Moves For 2027
- Move 1 — Keep Now Assist bundled in Pro Plus: never break it out as a per-user SKU. The ~30% uplift over Pro is the right vehicle because it sells to the *workflow buyer* (CIO, head of IT Service Management), not the *seat buyer* (CFO benchmarking against Copilot)
- Move 2 — AI Agent Studio on consumption pricing: per-execution and per-token meter, similar to Copilot Studio's message-pack model but priced on *agent runs* (an agent that opens an incident, gathers context across CMDB + HR + Finance, and resolves it = 1 high-value run worth $0.50-2.00, not $0.01/message)
- Move 3 — Workflow-outcome SKUs: "Now Assist for Customer Service" priced per-resolved-ticket (~$1-3/ticket), "Now Assist for HR" priced per-onboarded-employee (~$50-150/onboarding), "Now Assist for ITSM" priced per-incident-resolved. This is the moat — Microsoft has no equivalent because Copilot doesn't own the workflow record
The 1 Anti-Pattern To Avoid
- Do NOT ship a $30/user/mo Now Assist standalone SKU. This is the single highest-risk pricing move ServiceNow could make in FY27
- It invites apples-to-apples comparison with Copilot 365 — and Copilot wins on bundling math every time (because the customer already pays for M365)
- It cannibalizes the Pro Plus uplift, which is currently doing the heavy lifting on platform ARPU expansion
- Named precedent: Salesforce shipped Einstein 1 as a standalone $50/user/mo SKU in 2024 — attach rates were below 5% by mid-2025, and Salesforce quietly pivoted to Agentforce per-conversation pricing because the standalone SKU was getting CFO-shelved against Copilot
- Microsoft will price-match or undercut any per-user SKU within one quarter using EA discounting they can absorb and ServiceNow cannot
The Outcome-Pricing Pivot
- Per-resolved-ticket pricing: $1-3/ticket for ITSM/CSM, with a floor commitment — aligns ServiceNow margin with customer value, makes Copilot comparison nonsensical ("Copilot doesn't resolve your tickets")
- Per-onboarded-employee pricing: $50-150 per HR onboarding workflow completed end-to-end (laptop provisioned, accounts created, training assigned, manager notified) — this is a 6-12 hour manual process today
- Per-incident-resolved pricing: similar to ticket but for IR/SecOps workflows where the value-per-event is $500-5,000
- Named precedent — Intercom Fin AI: launched at $0.99 per resolution in 2023, hit ~$100M ARR by 2025 specifically because the pricing made AI ROI legible to buyers
- Named precedent — Salesforce Agentforce: $2/conversation pricing announced Sept 2024, explicitly modeled on Fin's per-resolution success after Einstein 1 standalone failed
- Named precedent — Zendesk AI agents: per-automated-resolution pricing launched 2024, similar trajectory — outcome pricing is now table stakes for the workflow-AI category
Sales Talk Tracks vs. Copilot
- Workflow context Microsoft can't match: "Copilot can summarize the email about the outage. Now Assist can read the email, open the P1 incident, page the on-call, pull the CMDB topology, and post the war-room link — all before the rep finishes reading the summary"
- Enterprise governance: Now Assist runs inside the customer's ServiceNow instance with existing RBAC, audit trail, and data-residency — Copilot for Service requires Dataverse + Power Platform governance setup separately
- Bundle math: "You're already paying for Pro. The Pro Plus uplift is ~30% — that's $7-17/user/mo equivalent for Now Assist *plus* AI Agent Studio access. Copilot 365 alone is $30/user, and it doesn't touch your CMDB"
- Named-customer reference patterns: lead with NVIDIA, Adobe, Deloitte Now Assist case studies — all Pro Plus customers running production workflows, not pilot/POC seats
- The CFO close: "Copilot is a productivity tax on every seat. Now Assist is a workflow tax on every *outcome*. One scales with headcount, one scales with value delivered"
Pricing Model Comparison
| Pricing Model | Today (May 2026) | Microsoft Copilot Comparable | FY27 Recommendation | Margin Profile | Risk |
|---|---|---|---|---|---|
| Per-user-per-month standalone | not offered | Copilot 365 $30/user | DO NOT SHIP | Low (race to bottom) | High (cannibalizes Pro Plus) |
| Tier uplift (Pro Plus) | ~30% over Pro, ~$7-17/user/mo equiv | none direct | Keep + expand | High (bundled margin) | Low |
| Consumption (AI Agent Studio) | rolling out | Copilot Studio msg-packs | Expand aggressively | Medium-High | Low (matches industry direction) |
| Outcome (per-resolved-ticket) | piloting in CSM | none | Launch GA in FY27 | Highest (value-aligned) | Medium (sales motion change) |
| Outcome (per-onboarded-employee) | not offered | none | Launch in HR Service Delivery | Highest | Medium |
| Per-tenant flat (Studio entry) | not offered | Copilot Studio $200/tenant | Optional dev tier | Low | Low (lead-gen) |
Pricing Strategy Decision Tree
FAQ
Why can't ServiceNow win a per-user price war with Copilot? Microsoft has roughly 400M paid M365 seats, so Copilot is incremental ARPU on an installed base with near-zero marginal distribution cost, and its inference is subsidized by Microsoft's own Azure capacity. ServiceNow is a renter of compute (NVIDIA H100s, partner clouds), not a hyperscaler, so it carries real infrastructure cost on every Now Assist execution.
How does Now Assist pricing actually map today? Now Assist is bundled in the Pro Plus tier at roughly a 30% uplift over Pro, which on a per-employee basis is about $80-200/employee/yr, or ~$7-17/user/mo equivalent when normalized. By comparison Copilot 365 runs about $30/user/mo on top of an M365 E3/E5 stack of $36-57, for a total of $66-87/user/mo.
What is the single anti-pattern to avoid? Do not ship a $30/user/mo standalone Now Assist SKU. It invites apples-to-apples comparison with Copilot 365, which wins on bundling math, and it cannibalizes the Pro Plus uplift McDermott told the Street is the FY26-27 growth lever. Salesforce's Einstein 1 standalone at $50/user/mo had sub-5% attach by mid-2025 before pivoting to Agentforce per-conversation pricing.
What are the three recommended pricing moves for 2027? Keep Now Assist bundled in the Pro Plus uplift, layer AI Agent Studio on top with consumption pricing (per-execution and per-token), and introduce workflow-outcome SKUs. The outcome SKUs include Now Assist for Customer Service priced per-resolved-ticket, Now Assist for HR priced per-onboarded-employee, and Now Assist for ITSM priced per-incident-resolved.
What pricing precedents support the outcome-pricing pivot? Intercom Fin AI launched at $0.99 per resolution in 2023 and hit roughly $100M ARR by 2025 by making AI ROI legible. Salesforce Agentforce's $2/conversation pricing, announced September 2024, was explicitly modeled on Fin's per-resolution success, and Zendesk AI agents adopted per-automated-resolution pricing in 2024 — making outcome pricing table stakes.
Bottom Line
Don't fight Microsoft on per-user price — you'll lose. Win on workflow context and outcome pricing, where Copilot has no equivalent. Keep Now Assist inside Pro Plus, layer Agent Studio consumption on top, and lead the category into per-resolved-ticket / per-onboarded-employee SKUs before Microsoft figures out workflow-AI is a different game than productivity-AI.
The standalone $30/user Now Assist SKU is the trap — Salesforce already walked into it with Einstein 1. (see also: q1616, q1620, q1631)