中文sales@inspirentech.co
Home/Insights/Why most companies still aren't getting value from AI, and how teams in Asia can measure it
Insights · 5 min read

Why most companies still aren't getting value from AI, and how teams in Asia can measure it

New surveys show few companies get real value from AI. Learn why AI efforts stall, what to measure, and a five-step way to prove AI value at work.

Almost every company now uses AI. Far fewer can say it is paying off. In October 2026, a Gartner survey of more than 2,000 CIOs found that only 13% report significant or transformational value from AI so far. The gap is rarely the technology. It is how teams choose, measure and stop AI work.

What does "AI value" actually mean?

AI value is a measurable improvement in a business outcome that you can trace back to how AI is used: hours saved on a task, faster turnaround for customers, fewer errors, more revenue per salesperson. It is not the number of licences bought, pilots launched or prompts typed. As Gartner analyst Alicia Mullery put it, "volume is not value."

What the latest surveys found

Three large surveys published in the second half of October 2026 point the same way:

SurveyWho was askedHeadline finding
Gartner, 2027 CIO Agenda2,002 CIOs and technology executives, May–July 202613% report significant or transformational value from AI. Only 31% have rules for when a pilot should not go into production, and 17% for shutting down low-value AI already in use.
Deloitte, State of AI in the Enterprise3,500 business and IT leaders in 25 countries, May–June 202681% report faster decisions and execution, but only 25% use AI to deeply transform the business and 28% report revenue gains.
MIT Technology Review Insights with Atlan250 executives including Asia-PacificEvery company uses AI, but 16% report innovative results and 23% see little or no effect relative to their investment.

The pattern: AI makes existing work faster, which is real, but most organisations have not changed the work itself or set up a way to prove the gain.

Why most AI efforts stall

  • No value defined up front. Teams start with a tool ("let's roll out Copilot") instead of a problem ("cut proposal turnaround from five days to two").
  • No baseline. If you never measured how long a task took before AI, you cannot show what changed.
  • Pilots that never end. Without clear criteria, weak pilots linger and strong ones never scale. Gartner found the CIOs getting the best results are 2.3 times more likely to have rules for stopping pilots.
  • Same process, new tool. Deloitte found most firms are speeding up existing workflows rather than redesigning them. Saving ten minutes on a report nobody reads adds little.
  • Missing context. AI that does not know your terms, rules and data definitions gives generic answers. Sales and finance may even read the same figure differently.
  • Untrained people. Staff who only know basic prompting use AI for small tasks and stop there.

A five-step way to measure AI value

  1. Pick one workflow, not one tool. Choose a task that is frequent, time-consuming and easy to check, such as drafting client proposals, summarising meeting notes or answering standard customer questions.
  2. Measure the baseline. For two weeks, record how long the task takes, how often it is redone and how many errors slip through.
  3. Set a target and a stop rule. For example: "Cut drafting time by 30% within eight weeks with no rise in errors. If we are below 10% at week eight, we stop or change approach."
  4. Train the people doing the work. Give them prompts, examples and checking rules for that exact task, using the tools you already have: Claude, ChatGPT, Microsoft Copilot or Google Gemini.
  5. Measure again, then decide. Compare against the baseline. Scale what works to the next team, fix what is close, and drop what is not delivering.

What to measure

Type of valueExample measures
TimeHours per task, turnaround time, time to first draft
QualityRework rate, errors caught in review, customer complaints
CapacityTasks completed per person, backlog size
RevenueProposals sent, win rate, response time to leads
CostAI licence and usage cost per workflow against the time saved
AdoptionShare of the team using AI for the task each week (a leading signal, not value by itself)

Keep it simple. One or two measures per workflow, tracked consistently, beat a dashboard nobody updates. Our free AI time savings calculator gives a quick first estimate.

Redesign the work, not just the task

The biggest gains come when teams ask whether the work still makes sense once AI is in it. Gartner found the best-performing CIOs are 2.4 times more likely to plan how roles will change when people work alongside AI, yet only 4% have such plans. In practice this can be modest: who reviews AI drafts, who approves them, which steps can be dropped, and which decisions must stay with a person. Our article on AI and human judgement at work covers where to keep people firmly in charge.

Keep governance in proportion

Deloitte found 23% of leaders say accountability is unclear when AI produces a bad outcome. Measurement and governance go together: name an owner for each AI workflow, set review rules that match the risk, and write them into a short policy. Our AI acceptable-use policy checklist is a practical starting point.

Where to start

Check where your team stands with the free AI readiness quiz, then use our 90-day AI adoption roadmap to plan your first measured pilot. For teams across Asia that want help choosing use cases and setting targets, our AI strategy and implementation workshop works through your own workflows.

Frequently asked questions

How do you measure the ROI of AI at work?

Pick one workflow, measure how it performs before AI, set a target, then measure again after training and use. Compare time, quality and cost against the baseline rather than counting licences or usage.

Why do most AI projects fail to deliver value?

Common reasons are no defined value up front, no baseline, pilots with no stop rule, applying AI to unchanged processes, and staff who have not been trained on the specific task.

How long before AI shows results?

Time savings on a well-chosen task often show within weeks. Revenue and deeper business change usually take longer, which is why setting a review point, such as eight weeks, helps.

What is a good first AI use case to measure?

A frequent, time-consuming task that is easy to check, such as drafting proposals, summarising meetings or answering standard customer questions.

This article is general information for business readers, not legal, financial or technical advice. AI tools and plans change quickly; check current vendor documentation before making decisions.

Keep exploring

Ready to make AI work for your team?

Tell us about your team and goals. We'll reply with a tailored training plan and quote.

Get a Quote