What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Companies can spend heavily on AI and still struggle to show that it improved the business. The gap is not proof that AI universally fails: it is the distance between investment or activity and outcomes an organization can measure, attribute to a use case, and repeat. Survey findings from 2026 point to weak measurement and organizational readiness as common challenges among the leaders surveyed.
What the reported numbers say—and whom they describe
HFS Research’s 2026 survey of 101 C-suite executives at enterprises with more than $1 billion in revenue found that 87% said their organization invests in AI faster than it can prove value. In the same survey, 72% lacked a consistent, trusted way to measure AI value, while 62% struggled to distinguish AI activity from real business results. These are respondents’ reports, not a census of companies. HFS Research’s report, produced in partnership with Wipro, also describes its sample as Fortune 200 firms in chart notes.
Confidence was limited: 21% of those surveyed were fully confident that their AI efforts represented measurable business value rather than signaling progress. Meanwhile, 65% said urgency and external pressure, rather than a clear plan, drove AI spending. The figures describe how the surveyed leaders viewed their organizations; they do not establish that every AI investment is wasteful.
A separate result offers another view, with a different source and scope. IBM’s 2026 enterprise cost-management guide reports that an IBM Institute for Business Value survey conducted with Oxford Economics found 37% of AI initiatives delivered the business value senior leaders expected by the end of 2025. That is a reported survey finding, not evidence that the same share applies to all AI initiatives worldwide. IBM’s guide is also a vendor publication, so its recommendations should be distinguished from the underlying survey result.
#1 Best Overall
Why activity is easier to see than business value
Launching a model, adding users, or counting generated outputs can show that a project is active. None, by itself, establishes that it reduced costs, increased revenue, improved service, or changed a business decision for the better. A credible result needs a defined use case, a baseline, an agreed target, and a way to connect the result and its costs to that use case.
HFS found that only 13% of surveyed organizations had AI deeply embedded in day-to-day workflows. It also reported that 83% of organizations in lightly contextual environments struggled to separate activity from outcomes, compared with 23% in deeply embedded environments. These are associations within the survey, not proof that deeper workflow embedding alone causes better results.
HFS’s report sums up its interpretation this way: “AI readiness is no longer primarily a technology challenge. The models are capable, but the operating models are not.” This is the report’s conclusion, not an independently tested universal finding.
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #2
Four questions that help explain the gap
Were success measures set before deployment?
Without a baseline and target, teams can report usage or output but cannot reliably establish whether the project improved on what came before. Define the outcome in business terms before rollout, then decide how and when to measure it. Suitable measures depend on the use case: IBM gives examples such as cycle-time reduction, cost avoidance, conversion lift, and faster incident resolution.
Does AI fit the workflow and have the right context?
A tool that operates outside the process it is meant to improve may produce activity without changing the work. Results depend on whether the system is connected to relevant organizational information and fits the decisions, handoffs, and constraints of the real workflow. A demonstration on generic inputs is not evidence that the system will perform usefully in a company’s operating environment.
Are people and decision rights in place?
Someone needs to own the business outcome, define what the system may decide or recommend, and determine when a person must review or override its output. Employee participation, training, and process redesign matter because a technically functional tool can remain peripheral if workers do not know how to use it or the surrounding process does not change.
Rank #3
Can a measured result scale beyond a pilot?
A promising pilot is a starting point, not proof of repeatable value. Before expanding it, assess whether the result holds across teams, locations, data conditions, and operating costs. Organizations also need a way to redesign or stop initiatives that miss agreed thresholds rather than treating continued spending as evidence of progress.
Recommended Free Tools
How to tell whether an AI investment is paying off
Compare each initiative on the same dimensions rather than relying on a single organization-wide AI score. IBM recommends cost attribution, outcome benchmarking, cross-functional governance, and ongoing portfolio optimization as parts of cost management.
- Define the business problem and owner. Name the workflow and the person accountable for its outcome. Specify what the AI system is intended to change, and who is responsible for acting on its output.
- Record a baseline and target. Measure the existing process before rollout. Set a target tied to a business outcome, plus a measurement period and a threshold for expanding, revising, or stopping the work.
- Connect the outcome to the use case. Choose a metric that reflects the intended change, such as cycle time, cost avoidance, conversion, or incident-resolution speed. Check whether other changes could explain the result rather than assuming AI produced it.
- Count the full cost of ownership. Include model and API fees, infrastructure, data pipelines, and engineering and data-science labor—not just compute invoices. Assign costs to the use case as consistently as possible so that the comparison is meaningful.
- Evaluate workflow integration and readiness. Check whether the system has the context it needs, fits into day-to-day work, and has trained users and clear human oversight. Record gaps that could limit adoption or performance.
- Review evidence before scaling. Compare actual results with the baseline and target over an appropriate period. Expand only when results and costs support the decision; otherwise, revise the workflow or stop the initiative.
The comparison should include total cost, the defined baseline and target, integration with workflow and context, ownership and readiness, the quality and time horizon of the evidence, and the ability to scale beyond a pilot. The sources do not establish a controlled ranking of interventions or prove that one approach reliably produces the best returns.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why potential value is not realized ROI
Large estimates of AI’s economic potential describe what might be affected, not what a company has earned or saved. A 2026 World Economic Forum article by Cognizant executives describes Cognizant analysis estimating that AI could affect $4.5 trillion worth of work in the United States today. That is a modeled potential-value estimate—not observed company revenue, savings, or profit. The article’s authors identify skills, contextual grounding, and designing around real business problems as ways to help translate capability into results; the WEF page says the views are the authors’ own.
For an individual company, investment pays off only when measured outcomes justify the full costs and the result can be sustained. The broader estimates and survey findings can frame the challenge, but they cannot substitute for evidence from the company’s own use cases.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
