80% Say AI Boosts Efficiency, But Profits Stay Flat — Why?

# 80% Say AI Boosts Efficiency, But Profits Stay Flat — Why?

Key Takeaways

McKinsey's State of AI in 2026 report, released August 25, finds that 80% of respondents say AI improved their personal efficiency, but only 37% say AI had a positive impact on company profits, and just 6% qualify as "AI high performers" (organizations where AI contributes at least 5% of EBIT). Personal experience and organizational results are two different ledgers — saved time does not automatically become profit. Where the gap comes from is the question every company must answer before adopting AI. Source: McKinsey 2026 global survey (May 4 – June 8, 2026, 1,719 responses across 97 countries, GDP-weighted). The 80% figure is self-reported perception; the 37% is corporate-level profit attribution, not an item-by-item audit.

What the Data Says

Adoption first: nearly 90% of organizations use AI regularly in at least one function, and 44% have scaled it enterprise-wide (up from 38% last year). AI agents reach 40% scaled adoption in large organizations (McKinsey 2026 global survey). Adoption speed is not the problem. Results second: 80% of people say their personal efficiency improved and about half say decision-making improved, but only 37% report a positive profit impact — roughly flat year over year. And the organizations actually making money from AI — those with AI contributing at least 5% of EBIT — account for just 6%. Put the two sets of numbers together and the conclusion is clear: AI has been adopted, individuals feel faster, but corporate profits have not grown.

Where the Saved Time Goes

Why do individual efficiency and organizational results diverge? Between "saved time" and "profit" lie four barriers. First, output never enters the business system: documents and plans AI produces are just files if they do not flow into real business processes. Second, old steps never exit: new processes go live while old ones keep running, and efficiency gains are cancelled out by duplicate work. Third, capacity has no destination: saved time gets refilled by new analysis, reviews, and meetings instead of becoming incremental output. Fourth, new costs are ignored: model fees, review time, rework, and maintenance are never counted. Miss any one barrier and you get the familiar outcome: individuals feel faster, and the organization is running an extra process.

What High Performers Do Differently

McKinsey calls organizations where AI contributes at least 5% of EBIT "high performers," and the difference is not the model — it is the way of working. Nearly three-quarters of high performers redesigned workflows because of AI, versus only about a quarter of other organizations. High performers are also more likely to show deep leadership involvement, impact measurement, and risk management. In plain language: ordinary organizations treat AI as an accelerator — the same processes, just faster. High performers treat AI as a restructuring tool — the process itself is redesigned. The value of AI is not in replacing a single action; it is in redesigning the entire chain.

What Companies Should Do Now

Stop chasing "efficiency" and think through three things first. One: define what happens after "faster." A task drops from one hour to ten minutes — where do the saved 50 minutes go? Is there an explicit arrangement? If you cannot answer, the efficiency gain is hollow. Two: verify with evidence, not feelings — ask five questions: Which task improved? Which business state changed? Where did the freed capacity go? What new costs appeared? Which business metric changed, and who confirmed it? A project counts only when all five have answers. Three: start by redesigning one process — do not stuff AI into an old workflow. Pick the most painful business line, rebuild the process, prove it, then replicate.

Verdict: AI Competition Enters the Second Half

AI competition in 2026 has moved from "are you using AI" in the past two years to "does AI deliver confirmable results." Leading companies are already pulling ahead through workflow redesign; the era of checking the box with a tool and a vague efficiency story is over. The real dividing line is not whether you buy AI — it is whether you are willing to redesign how work gets done around it. If you are, there is a place in the 6% high-performer club. If not, AI will only make processes more complex and costs higher.

FAQ

Q: Why did 80% personal efficiency gains not become profit?

A: Because saved time does not automatically become output. AI output not entering business systems, old steps not exiting, capacity not assigned, and new costs not counted — any missing link produces "individuals faster, organization not richer."

Q: How do I judge whether an AI project has value?

A: Use five layers of evidence: task improvement (baseline comparison), business state change (system records), capacity destination (old-step exit records), cost accounting (new expenses), and financial confirmation (which metric changed, who signed off). Evidence at all five layers means value.

Q: Can SMBs without big-model budgets still compete?

A: Yes. High performers are defined not by model size but by workflow redesign. SMBs can start with one painful business line, rebuild the process with mature AI tools, prove it, then replicate — this is the path most high performers took.

About Shanghai Naili Information Technology Co., Ltd.

Shanghai Naili Information Technology Co., Ltd. (Naili AI Logistics Lab) has focused on freight forwarder digital marketing for 10 years, providing full-service marketing operations, AI implementation, GEO optimization, and website transformation. Our default premise for forwarder AI projects is verifiability: every project is delivered against five layers of evidence — task improvement, business state change, capacity destination, cost accounting, and financial confirmation.