Kushe Bahl, Partner & Leader – Digital and AI Practice, McKinsey & Company urges leaders to move beyond hype and deliver measurable results.
At a recent keynote at 26th CIO&Leader Annual Conference , Kushe Bahl, Partner & Leader of the Digital and AI Practice, delivered a candid reality check: while artificial intelligence is flooding boardrooms with excitement, hard evidence of business impact is still scarce.
“We’ve talked about trillions of dollars of impact. Nobody has seen even a billion yet,” Bahl told the audience, stressing that the window to prove AI’s value is closing fast.
Excitement Meets Early Skepticism
Bahl compared the current moment in AI to the early days of digital transformation. Companies experimented aggressively, deployed dozens of tools, but struggled to prove outcomes. AI is following the same curve only much faster.
“Last year was all excitement. This year, skepticism is already creeping in and that’s before we’ve realized any real business value,” Bahl noted. Unlike digital, which took years before doubts emerged, the AI hype cycle is already showing cracks.
The 10% Rule: Where Impact Lies
One of Bahl’s most striking points was the “10% rule.” Most large enterprises can easily list 60 to 70 potential AI use cases. Yet only six or seven of them around 10% are capable of delivering transformational business impact, such as raising EBITDA by 5% or more.
The rest? Useful tools, good for saving time or improving workflows, but unlikely to move the financial needle. “It’s tempting to pursue the 90%,” Bahl warned. “But unless we bet on the tougher, high-impact use cases, we won’t see the business impact that boards and investors are expecting.”
Where AI Is Showing Promise
Bahl highlighted functional areas already demonstrating measurable gains:
Sales and Marketing: AI-driven lead generation and personalization delivering 3–4x better conversion rates.
- Customer Engagement: Smarter targeting and content creation driving 4–5x increases in interaction.
- R&D and Engineering: Faster time-to-market through simulation and data-driven testing.
- Customer Support: AI voice agents reducing costs by up to 40% in India.
- Software Development: Code-generation tools cutting new development costs by 40–50%.
But he cautioned that adoption matters as much as technology. For instance, AI-led lead qualification only works if sales teams commit to putting high-quality leads through the system not low-priority ones.
Risks, Governance, and Talent
Bahl also pointed to risks of bias, privacy, and hallucinations, warning that AI models often fail not at the edge cases, but on the “happy path” in ordinary use. Rigorous testing and training, he argued, must become new core capabilities.
He also highlighted the talent gap. While data science skills are abundant, what’s missing is strong product management the ability to design AI solutions that truly align with business needs. “We’ve given disproportionate importance to data science, and not enough to product management,” he stressed.
The Call to Action
Bahl’s closing message was direct: stop chasing breadth, and prove depth. “If each company here can deliver even one transformational case study by the end of the year, you’ll set benchmarks that the rest of the industry will follow,” he said.
His challenge reframes AI’s future: less about the number of pilots and more about the ability to show business outcomes that boards, CFOs, and shareholders can measure.