News analysis · Published

Workday's AI Sales: What Businesses Should Measure Before Scaling Agents

By the ELYMENT AI editorial team · Free to read

Workday reported on 27 August 2026 that AI generated more than 25% of its new annual contract value in fiscal Q2 2027 and that more than 5,500 customers use at least one of its organic agents. Those vendor-reported figures are a meaningful sign that enterprise AI is moving into software budgets, but they do not prove customer return on investment. Business leaders should measure active workflow use, accepted outcomes, cycle-time improvement, total operating cost and control failures before expanding an agent licence across the organisation.

A luminous enterprise AI workflow flowing through finance, HR and IT approval rails beneath the headline AI Agents Are Becoming a Budget Line.
Original ELYMENT.AI editorial illustration.

What Workday reported

Workday's fiscal second quarter ended on 31 July 2026. The company reported total revenue of US$2.649 billion, up 12.8% year on year, and subscription revenue of US$2.471 billion, up 13.9%. It said more than 5,500 customers now use one or more Workday-built agents, an increase of more than 35% from the previous quarter.

Chief executive Aneel Bhusri said AI contributed more than 25% of new annual contract value. Reuters separately reported that more than half of Workday's net-new wins during the quarter included at least one AI solution. Workday also highlighted products including Developer Agent, Agent Passport, Adaptive Decision Intelligence and its generally available Financial Audit Agent.

Why the adoption signal matters

Enterprise software buyers have spent much of the past two years testing whether AI is an add-on, a replacement threat or a reason to modernise a core platform. Workday's result suggests that AI can now influence the platform decision itself. Reuters cited analyst Rebecca Wettemann saying the AI proposition was strong enough to draw some customers into a platform switch.

That is commercially significant because HR and finance systems sit close to regulated data, payroll, audit evidence and material approvals. An agent adopted inside a system of record can become part of the operating model rather than a separate chatbot. It also increases switching cost if prompts, policies, evaluations and workflow history become tied to one vendor's platform.

Licence uptake is not business value

Workday's customer and contract metrics describe commercial adoption. They do not show how many licensed agents are active, how frequently they complete a useful task, how much human review remains or what customers save. The distinction matters because a product can appear in a contract before the organisation has changed a process or realised an outcome.

Operators should therefore avoid using licence count, generated content or tool calls as the main success measure. The useful unit is an accepted business outcome: a reconciled exception, completed audit package, approved forecast or resolved HR case delivered to the required standard. Cost should include implementation, integration, review, monitoring, remediation and vendor fees, not only the model or feature charge.

A five-metric agent value test

Before expanding an enterprise agent, compare a controlled baseline with production evidence across five measures.

  • Active adoption: the proportion of eligible users and workflows completing real work with the agent.
  • Accepted outcomes: the percentage of completed tasks that pass business and quality review without rework.
  • Cycle time: the end-to-end time saved after human review and exception handling are included.
  • Fully loaded cost: total platform, integration, monitoring, review and remediation cost per accepted outcome.
  • Control performance: policy breaches, approval overrides, incorrect actions, recoveries and unresolved exceptions.

What business leaders should do next

Ask vendors to separate contracted, enabled, monthly active and outcome-producing usage. Select one high-volume workflow with a stable baseline, define the accepted result and measure it for at least one complete operating cycle. Preserve your evaluation set, approval rules and evidence outside the vendor interface so a future platform decision remains possible.

ELYMENT AI's agent work brief provides a practical way to define the goal, sources, limits and approval conditions. Our agentic observability analysis explains why results need inspectable traces, while the cyber-insurance guide shows how permissions and incident evidence affect operational exposure. ELYMENT AI can help turn those measures into a repeatable business case before agent adoption becomes another unexamined software line item.

Sources

Continue learning

Frequently asked questions

How much of Workday's new sales came from AI in fiscal Q2 2027?

Workday said AI generated more than 25% of new annual contract value in the quarter. This is a vendor-reported sales metric, not a measure of customer ROI.

How many Workday customers use its AI agents?

Workday reported that more than 5,500 customers used at least one organic agent, up more than 35% from the previous quarter as of 27 August 2026.

What should businesses measure before scaling AI agents?

Measure active workflow adoption, accepted outcomes, end-to-end cycle time, fully loaded cost per accepted outcome and control or recovery failures.

Explore ELYMENT AI