Snowflake-Partnered Report Finds Only 48% of Agentic Analytics Insights Improve Business
A Snowflake-backed survey finds only 48% of agentic analytics insights improve business outcomes, while context, data and review gaps limit expected ROI.
Summary
Published November 2, 2026, a sponsored report produced in partnership with Snowflake finds that an average 48% of analytics-agent insights lead to actions that improve the business, up from 37% one year earlier. Organizations generally praise descriptive insights, but assessments of action-driving prescriptive insights are mixed. Among insight leaders, more than 70% of outputs improve the business, with their largest quality advantage in prescriptive analytics.
Insufficient business context is the leading quality problem, cited by 55% of surveyed executives, followed by fragmented or inaccessible data at 45%. Agents generate more insights than can be properly evaluated at 52% of organizations; only 24% evaluate every output, while most rely on human-intensive, case-by-case review. Companies are prioritizing stronger AI and data governance, cited by 60%, and semantic data layers, cited by 45% and described by interviewed experts as crucial. Poor outputs can undermine returns or trigger financial and reputational damage. The report says humans conducted the research and writing, with any AI use limited to production processes under human oversight.
Positives
- 48% of agent insights now produce business-improving actions, up from 37% one year earlier.
- More than 70% of insights improve the business among the report’s insight leaders.
- 60% of surveyed companies are prioritizing stronger AI and data governance.
- 45% are adding semantic layers that interviewed experts consider crucial to high-performing agents.
Risks & concerns
- 52% of organizations receive too many agent insights to evaluate and act upon properly.
- Only 24% of surveyed organizations can evaluate every agent-generated insight.
- 55% of executives cite insufficient business context as the leading cause of poor insight quality.
- 45% blame fragmented or inaccessible data, while flawed outputs can cause financial or reputational damage.
- Prescriptive insights receive mixed reviews despite their role in determining automated actions.