Fewer than half of insights generated by analytics agents at the surveyed organizations lead to business improvement, according to a report from Insights, MIT Technology Review's custom content arm. The report says 48% of agent insights lead to actions that improve the business, up from 37% one year earlier.
The report finds that prescriptive insights, the ones that trigger actions, receive mixed reviews from respondents, while descriptive insights draw more praise. Flawed insights that prompt incorrect actions can cause reputational or financial damage, the report says.
A small group of organizations the report calls "insight leaders" reports that more than 70% of agent insights lead to beneficial actions. The report says the quality gap between these leaders and the wider group is widest for prescriptive insights.
Executives surveyed cite a lack of business context as the main cause of substandard insights, at 55%. Fragmented or inaccessible data follows at 45%. The report says fewer insight leaders report struggles with access to context.
The report also says 52% of organizations generate more insights than they can properly evaluate. Only 24% evaluate all agent insights, and most review them case by case in a process the report describes as human-intensive.
To improve insight quality, surveyed companies prioritize strengthening AI and data governance (60%) and adding a semantic layer to their data architecture (45%). The report says the experts interviewed consider the semantic layer crucial for high-performing agents.