The number that should worry every CIO
Accenture's latest Pulse of Change report, which surveyed 3,000 C-suite leaders and 3,000 non-C-suite employees, found that only 23 percent of companies reported measurable business value from their AI investments in July 2026. That is down from 32 percent earlier in the year, a nine point decline at a moment when AI spending across the enterprise has continued climbing rather than pulling back. Value realization is moving in the opposite direction from investment, and that divergence is the headline finding here, not the productivity gains that usually dominate reports like this one and give boards an easy, comfortable story to tell.
The decline is notable because it is not explained by the technology underdelivering at the individual level. More than two thirds of C-suite leaders said agentic AI exceeded their productivity expectations, and a similar share of employees reported greater job satisfaction working alongside these tools day to day. Something is happening in the space between individual productivity gains and enterprise-level, board-reportable value, and Accenture's data suggests that gap is widening rather than closing as adoption matures across more teams and more use cases.
Why productivity gains aren't showing up as business value
Muqsit Ashraf, Accenture's global lead for industry and enterprise, put the diagnosis plainly: companies have rolled out tools and copilots that produce local productivity but not enterprise-level impact. That distinction matters because it describes exactly the failure mode most CIOs will recognize from their own AI rollouts. A team saves hours a week using a copilot, everyone feels good about it, and none of that time savings ever converts into a metric a CFO or board member would recognize as value: revenue, cost reduction, or cycle time on a process that matters to the business.
Ashraf's prescription is concentration over breadth: take a handful of agentic use cases all the way to a reportable outcome, rather than running hundreds of shallow pilots that each generate anecdotal wins nobody can aggregate. His sharpest line frames the stakes directly: AI has been assistive so far, but an organization does not transform its business by doing old things faster. That is a real challenge to the pilot-heavy strategy most enterprises have defaulted to, where breadth of experimentation gets mistaken for progress.
The workforce data underneath the value gap
The report's workforce numbers add texture to the value problem. Seventy eight percent of leaders expect employee roles to change within a year, and 57 percent of employees say their roles are already changing, which suggests the disruption is real and underway regardless of whether it shows up on a P&L yet. Seventy three percent of leaders expect entirely new AI-focused entry-level positions to emerge within a few years, pointing to structural workforce change that outruns the measurement systems currently in place to track its value.
One number stands out as a plausible root cause: only 25 percent of top executives use AI daily themselves. Leaders who are not hands-on with the tools their organizations are deploying are poorly positioned to identify which use cases are worth scaling versus which are producing the kind of local, unmeasurable productivity Ashraf describes. Executive AI literacy is not a soft skills issue in this context. It directly affects whether leadership can tell the difference between a pilot worth doubling down on and one worth killing.
Why investment keeps rising despite falling measured value
Despite the decline in reported value, 82 percent of C-suite leaders plan to increase AI investment going forward. That is not necessarily irrational: it can reflect genuine confidence that the technology works, paired with an honest admission that the organization has not yet figured out how to capture that value at scale. But it is also exactly the pattern that produces budget scrutiny later, when a board asks why AI spending keeps growing while the value metrics tracking it are flat or declining.
For a CIO, this is the moment to get ahead of that scrutiny rather than wait for it. Accenture's data gives a defensible narrative: productivity and satisfaction gains are real and measurable at the individual level, and the current gap is a portfolio management problem, not a technology failure. That narrative only holds up, though, if the CIO can point to a small number of use cases that have been taken through to a reportable business outcome. Without that proof point, rising investment against falling measured value is a hard story to defend in front of a skeptical board that reads the same survey data everyone else does.
The portfolio discipline this data demands
The practical shift Accenture's report argues for is moving from a pilot-counting mentality to an outcome-counting one. Most enterprise AI programs today report progress in terms of pilots launched, tools deployed, or users onboarded, all of which are activity metrics rather than value metrics. Ashraf's recommendation, to take a handful of use cases all the way to a reportable outcome, requires killing pilots that are not on a credible path to measurable value, which is organizationally harder than launching new ones.
CIOs building their next budget cycle around this data should expect the conversation with finance and the board to shift accordingly. The defensible position going into 2027 planning is not the number of AI initiatives underway but the number that have crossed into measured business value, however small that number currently is. Accenture's 23 percent figure is a low bar, but it is the honest bar, and it is the one CIOs should be prepared to beat rather than obscure with activity counts.
The retail and SaaS read on this data
For PE-backed SaaS and retail technology leaders in particular, Accenture's finding lands close to home because these sectors have been among the most aggressive adopters of copilots and agentic tools inside customer service, merchandising, and engineering workflows. That aggressiveness is exactly why the value gap matters more here than in slower-moving industries: a portfolio company burning cash on dozens of AI pilots without a reportable outcome to show a board or a potential acquirer is accumulating exactly the kind of unmeasured spend that shows up as a red flag in due diligence.
The fix Ashraf is describing, concentrating effort on a handful of use cases taken to completion, maps cleanly onto how PE operating partners already think about portfolio triage across a stable of companies. Treat the current slate of AI pilots the same way a portfolio review treats underperforming initiatives: keep the ones with a credible line to a P&L metric, and sunset the rest before the next board meeting rather than after it, when a partner is asking the questions instead of the CIO.
