Enterprise AI Adoption Faces Major Challenges: Data Quality, Integration, and Governance Issues, ETCIO

Share with your CIO

Most large enterprises are stuck in AI’s awkward middle phase, and a UiPath survey of nearly 600 C-suite and IT leaders at companies above $1 billion in revenue puts numbers to that stall. Only 31% say AI is fully embedded across their organisations. Data quality leads the obstacle list at 38%, workflow integration sits just behind at 37%, and governance trails at 33%. The survey’s sharpest finding is the ROI gap it implies: among organisations where orchestration is fully embedded, 89% say their agentic AI deployments met or exceeded return expectations.

What this means for your business

The 89% ROI figure is the number worth sitting with, because it reframes the entire deployment debate. It suggests the bottleneck isn’t the AI model itself but the orchestration layer sitting beneath it, the connective tissue that routes agents across systems, enforces rules, and makes outputs visible to the humans nominally in charge. If your organisation is in the 71% that haven’t fully embedded orchestration, the ROI underperformance you’re seeing isn’t a model problem. It’s an architecture problem, and model upgrades won’t fix it.

The survey comes from UiPath, a company whose core product is process automation and orchestration, which tilts the framing toward orchestration as the master variable. That’s worth noting, but it doesn’t invalidate the finding. The data quality and integration numbers, 38% and 37% respectively, align closely with what independent research has consistently shown. The orchestration-ROI correlation is harder to verify without the raw methodology, yet it describes a plausible mechanism: agents operating without orchestration produce outputs that can’t be trusted, audited, or acted on at scale, which is exactly why pilots stay pilots.

The decision this reframes isn’t whether to invest in agentic AI. Most CIOs reading this have already made that call. The question is sequencing. Organisations that treat data readiness and orchestration as infrastructure to fund before expanding agent scope are building a foundation the ROI numbers can compound on. Organisations that keep layering pilot use cases onto unresolved integration debt are essentially adding floors to a building with a cracked foundation. If your next budget cycle doesn’t include a line for orchestration infrastructure and data governance, that’s the budget worth defending differently.

Based on reporting from Enterprise AI Adoption Faces Major Challenges: Data Quality, Integration, and Governance Issues, ETCIO, originally published 2026-09-25 22:30:00.