McKinsey’s annual AI report has become required reading for enterprise leaders. The 2026 edition continues patterns of useful data mixed with consulting-firm positioning. Here’s how to extract value while maintaining critical perspective.

What’s Useful

Enterprise Adoption Data

McKinsey surveys hundreds of enterprises about AI adoption. This produces genuinely useful data:

Finding: 72% of enterprises report using AI in at least one business function (up from 55% in 2024).

Why it’s useful: Large-sample survey data provides benchmarking context. You can assess how your organisation compares to peers.

Caveat: Self-reported “AI use” varies in meaning. Some organisations count Copilot deployment; others only count custom AI. Comparisons require understanding definitions.

Spend Distribution Patterns

Data on how AI budgets are allocated across categories provides planning context:

Finding: Productivity AI (42%), custom applications (28%), infrastructure (18%), capability building (12%).

Why it’s useful: Budget allocation benchmarks help finance conversations and identify where you might be over or under-investing.

Caveat: These are averages. Your appropriate allocation depends on your specific situation, maturity, and priorities.

Barrier Identification

Survey data on what organisations cite as AI adoption barriers:

Finding: Data quality (68%), talent (64%), integration complexity (52%), unclear ROI (48%), governance concerns (41%).

Why it’s useful: Confirms that common challenges are actually common. You’re not alone in facing them.

Caveat: Self-reported barriers may reflect convenient excuses rather than actual blockers. Data quality is easy to blame; internal politics is harder to admit.

What’s Vendor-Influenced

Technology Predictions

McKinsey’s technology predictions often align with major vendor roadmaps. This isn’t coincidence – consultancies have deep vendor relationships.

Example claim: “Agentic AI will transform knowledge work by 2028.”

Critical read: This aligns with major vendor positioning (Microsoft, OpenAI, Google). Agent capabilities remain limited in production. The prediction may reflect vendor hopes more than demonstrated trajectory.

How to use: Note the prediction but weight your own experience and evidence more heavily than consulting firm projections.

Productivity Claims

Reports typically cite dramatic productivity improvements from AI adoption.

Example claim: “AI adopters report 40% productivity improvement in applicable tasks.”

Critical read: Self-reported productivity improvements are unreliable. Respondents have incentives to overstate success. “Applicable tasks” does much work in this sentence.

How to use: Treat productivity claims as upper bounds, not expectations. Your results will likely be lower.

Urgency Framing

Consulting reports often emphasise urgency – competitors are moving fast, you risk being left behind.

Example framing: “Leaders are pulling ahead. Organisations that delay AI adoption risk permanent competitive disadvantage.”

Critical read: Urgency framing serves consulting sales. It creates pressure that drives engagement. The reality is more nuanced – some organisations benefit from fast-follower strategies.

How to use: Consider urgency in your specific competitive context, not generically.

How Consulting Firm Research Works

Understanding research methodology helps critical reading:

Survey Methodology

Large consulting firms survey their client base – which skews toward larger enterprises that can afford consulting fees. Results may not represent the broader market.

Analyst Incentives

Research supports consulting sales. Reports that say “this is hard and you need help” generate more revenue than “this is straightforward and you can do it yourself.”

Vendor Relationships

Major consultancies partner with technology vendors. Research that validates vendor positioning strengthens those relationships.

Publication Bias

Successful case studies get published. Failures don’t. The reported success rate of AI projects likely exceeds actual success rate.

None of this makes the research valueless – but understanding the context improves how you use it.

How to Read Consulting Firm AI Research

Extract the Data, Question the Interpretation

Survey data, spending figures, and adoption rates are useful inputs. Interpretation and recommendations deserve more scepticism.

Cross-Reference Multiple Sources

Compare McKinsey findings with Gartner, Forrester, academic research, and your own experience. Where sources agree, have more confidence. Where they diverge, investigate why.

Consider the Incentives

Ask: who benefits from this conclusion? If the conclusion is “you need expensive consulting help,” apply appropriate scepticism.

Apply to Your Context

Generic findings may not apply to your specific industry, size, or situation. Extract principles, not prescriptions.

Weight Experience Over Projections

Consulting projections about future technology capabilities are often wrong. Your own experience with current technology is more reliable than projections about future technology.

The Valuable Practice

Despite limitations, consulting firm AI research has genuine value:

Benchmarking context: Understanding what peers are doing helps calibrate your own efforts.

Executive communication: Credible third-party data supports internal business cases.

Trend identification: Aggregate patterns across enterprises reveal emerging themes.

Discussion catalyst: Research reports provide common reference points for strategic conversations.

The practice is using this value while maintaining critical perspective on limitations.

Final Thought

McKinsey’s AI research contains useful data within a commercial context. The data is worth extracting; the urgency and positioning are worth questioning.

Read consulting firm research as one input among many. Weight your own experience, peer conversations, and direct observation at least as heavily.

The enterprises that succeed with AI are those that think for themselves – using external research as input, not as substitute for judgment.