McKinsey says only 7% of organizations have fully deployed AI. The problem may have less to do with technology and more to do with how companies operate.
Companies have spent billions putting AI to work. It writes emails, summarizes meetings, scores leads, and generates forecasts. Sales teams have more automation than ever, yet many are still struggling with the same old problems: missed quotas, unreliable forecasts, and customer information scattered across half a dozen systems.
Apparently, giving everyone an AI assistant doesn’t automatically make the company smarter.
A new McKinsey report on AI in banking offers an explanation. The research focuses on distressed credit management, but its conclusions should get the attention of every CRO and RevOps leader. In other words, AI adoption is accelerating. Operational transformation isn’t.
According to McKinsey’s 2026 State of AI research, fewer than one-third of organizations have started scaling AI across the enterprise. Just 7% have fully deployed the technology. The other 93% have some work to do.
The AI Pilot Problem
McKinsey found that banks are using AI agents to handle customer conversations, recommend payment terms, retrieve policies, and automate administrative tasks. The technology works. Productivity improves. Customers get faster assistance. But here’s the catch: Many banks haven’t fundamentally changed how their operations function.
Sound familiar?
In sales organizations, marketing uses AI to identify prospects. Sales uses another tool to prioritize accounts. Customer success tracks retention risks somewhere else. Finance calculates commissions using data that may not match anyone else’s numbers.
Each department becomes more efficient at doing its own work. Meanwhile, the handoffs between departments remain a mess. That’s the distinction McKinsey makes between AI assistance and AI orchestration. Assistance makes individual tasks easier. Orchestration connects those tasks into a coordinated process that produces measurable business outcomes. One saves time. The other can change how a company makes money.
What Banks Understand About AI That Sales Teams Should Steal
McKinsey describes an operating model in which specialized AI agents coordinate work across the customer lifecycle. Rather than waiting for an employee to identify a problem, agents continuously analyze information, prioritize actions, route work, and recommend next steps.
Consider how that approach could work in revenue operations.
Territory planning: Instead of reviewing territory assignments once or twice a year, AI could continuously identify coverage gaps, workload imbalances, and changes in market opportunity.
Lead routing: Rather than relying entirely on static assignment rules, intelligent routing could account for account ownership, rep capacity, customer history, and urgency.
Revenue intelligence: Customer conversations could inform pipeline health, forecasting, coaching, and retention strategies without requiring someone to manually transfer information between systems.
Sales compensation: Commission calculations could stay connected to territory changes, quota adjustments, and actual sales performance.
None of these capabilities is particularly revolutionary on its own. The opportunity comes from connecting them.
A territory change should inform routing. Routing should support pipeline execution. Pipeline performance should inform forecasting. And compensation should reflect the plan everyone agreed to in the first place. That’s a revenue operating model, not another collection of AI features.
The Human Judgment Question
McKinsey also makes an important distinction about autonomy. Not every decision belongs to an AI agent. Banks can automate routine customer communications and case assignments while reserving sensitive financial decisions for human review.
Revenue organizations face similar choices. AI can identify an underperforming territory. But should it automatically reassign strategic accounts? It can recommend quota adjustments. But should it make compensation decisions without leadership approval?
Probably not.
The goal isn’t to remove people from revenue operations. It’s to eliminate repetitive work while giving people better information for decisions that matter. Good orchestration requires clear decision rights, reliable data, and governance. Otherwise, companies risk automating confusion at a much higher speed.
Stop Measuring AI Activity. Start Measuring Revenue Outcomes.
McKinsey recommends that banking executives look beyond pilot performance and measure improvements in customer satisfaction, productivity, and financial results. Are AI investments improving win rates? Are leads reaching the right sellers faster? Are forecasts becoming more accurate? Are territories better balanced? Are commission disputes decreasing?
An AI tool that saves a salesperson 30 minutes is useful. But a connected operating model that helps the organization capture revenue it would otherwise lose is considerably more valuable.
This is where Fullcast’s Plan-to-Pay approach becomes relevant. Connecting territory and quota planning, routing, forecasting, revenue intelligence, and commissions creates the operational foundation needed to turn isolated improvements into coordinated execution.
AI can strengthen that foundation. It can’t substitute for it.
The Next AI Advantage Won’t Come From Another Tool
McKinsey’s central argument is that the organizations making the biggest gains won’t necessarily have the most sophisticated models or the largest collection of AI pilots. They’ll be the ones willing to redesign how work gets done.
For RevOps leaders, that means examining the entire revenue lifecycle instead of buying another solution for whichever department complains the loudest.
The technology is moving quickly. The operating model needs to catch up. Because making every department faster doesn’t help much when they’re all running in different directions.
4 Key Takeaways
1. AI adoption isn’t the same as AI transformation. McKinsey reports that only 7% of organizations have fully deployed AI. Companies are investing in smarter technology without fundamentally changing how work gets done.
2. AI assistance improves tasks. AI orchestration improves outcomes. Automating emails, lead scoring, and forecasting saves time, but connecting those activities across the revenue lifecycle creates opportunities for measurable business growth.
3. Disconnected RevOps systems undermine AI investments. Territory planning, lead routing, forecasting, and commissions must work together. Automating isolated processes only makes disconnected departments faster at operating independently.
4. Measure AI success by revenue results, not productivity gains. Faster workflows mean little without improvements in win rates, forecast accuracy, lead response times, and revenue attainment. The real competitive advantage comes from redesigning operations around measurable outcomes.