Why 94% of Companies See No ROI From AI Spending

Why are companies spending record amounts on AI but not moving earnings?

Because productivity gains and earnings impact are two completely different things. Workers feeling more productive is real. That productivity translating into margin, revenue, or cost reduction that shows up in financials is a different problem entirely, and most organizations have not solved it.

McKinsey's 2026 State of AI survey, which polled 1,719 executives, put a number on what a lot of operators already suspected: 94% of companies cannot attribute significant earnings impact to their AI investments. Meanwhile, 80% of individual workers report productivity gains from AI tools. That gap tells you everything about where the problem lives.

If you're a small or mid-size business operator watching enterprises pour money into AI with nothing to show for it, this should be clarifying, not discouraging. The failure mode is predictable. That means it's avoidable.

What does the 6% actually do differently?

The companies seeing real earnings impact share a few structural behaviors that separate them from the majority. These aren't secret tactics. They're discipline decisions that most leadership teams skip because they're harder than buying software.

They connect AI deployments directly to a financial metric before they build anything.

Not "we want to improve efficiency." Not "we want to be an AI-forward company." Something specific: reduce cost-per-ticket by 18%, cut quote turnaround from 4 days to same-day, recover 6 hours per week per sales rep that goes back into pipeline. The deployment gets scoped around that number, and success is measured against it.

They redesign workflows instead of layering AI on top of broken ones.

This is where most ROI dies. A company buys an AI writing tool and drops it into a content process that already had four approval layers and unclear ownership. The tool saves 30 minutes of drafting and adds 2 hours of confusion. Net result: negative. The organizations seeing earnings impact treat an AI deployment as a reason to audit and simplify the workflow first, then automate the clean version.

They assign ownership of the outcome, not the tool.

Someone is responsible for whether the metric moves. Not for whether employees are using the software, not for whether the tool is configured correctly, but for whether the number changes. That accountability structure is uncommon and uncomfortable, which is exactly why most companies avoid it.

Why does individual productivity not translate to earnings?

This is the core paradox in McKinsey's data and it deserves a direct answer.

When a worker saves an hour using an AI tool, that hour goes somewhere. In most organizations, it gets absorbed into the existing workload. More Slack messages answered, more meetings attended, more tasks that were previously deprioritized. The hour disappears into organizational slack rather than being redirected into something that generates revenue or reduces cost.

80% of workers report productivity gains. 6% of organizations see earnings impact. The difference is what happens to the recovered time and capacity.

This is not an AI problem. It's a management problem. Organizations that capture the value of AI productivity gains make an explicit decision about where the recovered capacity goes. That decision has to come from leadership, and it has to be enforced.

How does company size affect AI ROI outcomes?

Here's where the McKinsey data has a practical implication for SMB operators that most enterprise-focused coverage misses.

Large enterprises have a specific liability: they have more surface area for AI tools to spread without coordination, more organizational layers to absorb productivity gains before they reach the P&L, and more political complexity around workflow redesign. The 94% failure rate is, in part, an enterprise-scale coordination failure.

Smaller businesses have a structural advantage. If you run a 20-person company and you redesign a workflow around an AI tool, you can see the impact in 30 days. You don't need a change management program or an enterprise rollout. You change the process, you measure the output, you adjust.

The SMB operator who takes the same disciplined approach as the top 6% of enterprises, connecting deployments to specific financial metrics, redesigning workflows before automating, assigning outcome ownership, can move faster and see results more clearly than any Fortune 500 running the same playbook.

What are the most common ways AI deployments fail to produce ROI?

Based on what we see working with clients and what McKinsey's data confirms, the failure patterns are consistent.

| Failure mode | What it looks like | Why it kills ROI | |---|---|---| | Tool-first deployment | "We bought Copilot. Now what?" | No metric, no workflow change, no ownership | | Productivity theater | Workers feel faster, output stays flat | Recovered time absorbed into organizational slack | | Layering on broken workflows | AI added to a process that was already inefficient | Automates the dysfunction instead of fixing it | | No outcome owner | IT owns the tool, nobody owns the result | Nobody is accountable when the metric doesn't move | | Metric mismatch | Measuring usage instead of financial impact | You optimize for adoption, not earnings |

Every one of these is a structural choice, not a technology limitation. The AI tools work. The question is whether the organization around them is set up to capture the value.

What we'd actually do

• Pick one process, one metric, and build backward. Before any new AI deployment, write down the specific financial number you expect to move (cost, revenue, time that converts to capacity with a defined use). If you can't write that down, don't deploy yet.
• Audit the workflow before you automate it. Map the current process. Remove steps that only exist because of old constraints. Then automate the clean version. Automating a broken workflow just breaks things faster.
• Assign a result owner, not a tool owner. Someone on your team is responsible for whether the metric moves by a specific date. That person has authority to change the process and is accountable for the outcome. This single structural decision separates the 6% from the 94%.