Survey of 2,025 Agentic AI Leaders Finds Data Readiness, Not Launch Speed, Drives ROI

Salesforce published survey results timed to its Dreamforce 2026 conference (September 15-18 in San Francisco) polling 2,025 agentic AI decision-makers across 20 countries between May 14 and May 28, 2026. The headline finding cuts against the industry's move-fast instinct: organizations that unified their data before deploying AI agents reached measurable ROI in an average of 7.3 months, versus 8.8 months for the 69% of respondents who deployed agents against fragmented or incomplete data first. Across the full sample, average time to meaningful ROI was about eight months, with 30% of respondents already in production, 47% piloting, and 23% still evaluating.

The industry breakdown complicates the assumption that technical sophistication predicts speed. Professional and business services firms reached ROI fastest, in 6.5 months, followed by engineering, manufacturing, and financial services at roughly 7.2 months. High tech companies, despite presumably deeper AI fluency, were slowest at 10.1 months, according to Salesforce's report. The survey also tied governance intensity to outcomes: respondents with above-average governance structures took longer to reach ROI (9.3 months versus 7.2 months) but caught errors before they reached customers far more often, 18% of the time versus 32% for lighter-governance deployments.

The report's centerpiece customer story was Asymbl, a Salesforce-aligned services firm that built a digital recruiting agent, nicknamed Rosa, inside Slack. Independent Dreamforce coverage from SiliconANGLE reported Asymbl used the agent to hire roughly 100 people in 100 days, that digital workers now account for about half of staff across 13 functions, and that the company's measured productivity value from its AI program grew from $5 million to $13 million year over year. Salesforce's own report separately cited a 96% retention rate among hires sourced through the agent.

"The advantage was never in starting first; it's in starting deliberately." — Shibani Ahuja, SVP, Data & AI Strategy, Salesforce

Independent reporting on the same conference week complicated the narrative Salesforce was telling. MarketScale's Dreamforce coverage noted that the event's own pre-show materials included no customer-reported ROI, adoption metrics, or cost-to-run figures for Agentforce itself, the flagship product the survey and case studies were meant to validate. Coverage from theCUBE Research analyst George Gilbert, cited by SiliconANGLE, credited Salesforce with unusually tight integration across its agent stack but stopped short of endorsing any industry-wide payback figure, and Salesforce's own survey data is explicitly self-reported by respondents rather than independently audited.

Taken together, the two threads point to where enterprise AI adoption actually stands roughly two years into the agent era: a maturing playbook around sequencing, data unification, and scoped deployment that credibly predicts faster payback within any given company, alongside a persistent absence of standardized, third-party-verified benchmarks that would let one company's results be compared against another's. For buyers evaluating their own rollouts, the survey's sequencing advice is more actionable than its headline ROI timeline, precisely because the timeline still rests on vendor-selected case studies rather than audited, product-wide numbers.