The consulting giants just quantified what small businesses have been living — and the numbers explain why your competitors are pulling ahead.
A client came to me last year convinced they needed AI for social media content creation. Their team was drowning, they said. Content couldn’t keep up with demand.
I ran an audit. Content creation took two hours a week. Content approval and posting? Eight hours.
They’d spent six months optimizing the wrong problem. The AI tools they bought worked perfectly. The workflow around those tools was the failure. I wrote about this pattern when I first started seeing it with clients — the death spiral where better tools actually make broken processes worse.
Last month, PwC put a number on what I saw on that Google Meet.
Technology Delivers 20% of the Value. The Other 80% Is Your Problem.
In their 2026 AI Business Predictions, PwC states it plainly: “Technology delivers only about 20% of an initiative’s value. The other 80% comes from redesigning work.”
Let that land for a second. Four out of five dollars of AI value have nothing to do with the AI.
Deloitte’s 2026 State of AI in the Enterprise report, published two weeks ago, tells the same story from the other direction: only 34% of organizations are truly reimagining their business with AI. Meanwhile, 37% are using AI at a surface level with no changes to underlying processes. They bolted a jet engine onto a bicycle and are wondering why they’re not flying.
It gets worse. Deloitte’s workforce research found that 84% of organizations haven’t redesigned roles based on AI capabilities. Only 16% have fully designed roles, processes, and operating models to integrate AI. Deloitte’s own researchers put it bluntly: “If you just take your existing workflow and try to apply advanced AI to it, you’re going to weaponize inefficiency.”
The $2.5 Trillion Misdiagnosis
Here’s the pattern I keep seeing — at the enterprise level and in small businesses alike.
Companies buy AI tools. They layer them onto existing workflows. Results disappoint. So they buy more AI tools. I called this the automation budget problem last October — organizations spending aggressively on technology while their actual bottlenecks sit untouched.
PwC’s own CEO survey confirms the scale of this misdiagnosis: fewer than 25% of CEOs say AI is applied extensively across core activities. Most report no meaningful revenue gains or cost reductions from AI investments in the past year.
And PwC identifies exactly why. Companies default to what they call “crowdsourcing AI efforts” — encouraging bottom-up adoption, letting teams experiment freely, hoping it all adds up. The finding? It “seldom produces meaningful business outcomes.” Impressive adoption numbers. Minimal business impact. Think about that the next time someone brags about their company’s AI adoption rate.
I had a manufacturing client who lived this. The team adopted ChatGPT, Jasper, Copy.ai, and Canva AI. Software costs tripled. They produced four times more content. The marketing manager went from leaving at 5:30 to working until 8 PM managing tools. The owner told me: “We have more content than ever, but I can’t tell you if any of it’s actually working.”
That’s the 80% in action. Or rather, the 80% being ignored.
Why Small Businesses Are Accidentally Winning This Game
Here’s where the data gets interesting — and where I think most AI coverage gets the story completely wrong.
While enterprises are stuck in the gap between AI investment and AI value, small businesses using AI to scale are reporting dramatically different results: 93% saw revenue grow, 82% reduced costs, and 91% reported year-over-year ROI. Salesforce found similar numbers — 91% of SMBs with AI say it boosted revenue, and 86% report improved profit margins.
Compare that to enterprise, where only 20% have seen the revenue growth they expected from AI.
The gap isn’t budget. It isn’t technical sophistication. It’s structural freedom.
I’ve written about this advantage before — the idea that small businesses have architectural flexibility that enterprises can’t match. When you’re a five-person company, you can’t “layer AI onto legacy processes” because you don’t have legacy processes. You build the workflow around the tool because there’s nothing else to build around. The 80% — redesigning work — happens by default, not by initiative.
When a large enterprise tries to redesign a process, they face change management committees, stakeholder alignment, IT approval cycles, and the gravitational pull of “that’s how we’ve always done it.” When a small business owner decides to redesign a process, they redesign it. Sometimes before lunch.
PwC calls this the “rise of the generalist” — someone who understands a range of tasks well enough to oversee AI agents working across them. That’s literally every SMB owner I’ve ever met. You’re already the generalist. You’ve been the generalist since day one. The future of small AI isn’t about catching up to enterprise. It’s about recognizing you were already ahead.
What “Redesigning Work” Actually Looks Like Without a McKinsey Engagement
Here’s the practical part, because data without action is just trivia.
Before you buy, subscribe to, or even trial any AI tool, ask yourself one question: can you draw the workflow it’s going into — every step, every handoff, every approval?
If you can’t draw it, you can’t improve it. My content creation client would have found the eight-hour approval bottleneck in twenty minutes with a whiteboard. They skipped the twenty minutes and wasted six months.
The second move: kill the handoff, not the task. Most workflow waste isn’t in the tasks themselves — it’s in the waiting, the approvals, the “can you send me that again in a different format.” AI doesn’t fix the handoff. It often makes it worse by generating more output that feeds into the same broken handoff. Better operations will always beat better agents.
The third move: measure the whole chain. “Faster content creation” isn’t a metric. “Same quality content with 40% less total time from concept to publication” is. Deloitte found that 74% of organizations want AI to drive revenue growth, but only 20% have seen it — and a major reason is that they’re measuring tool performance, not workflow performance. Gartner projects that 40% of agentic AI projects will fail by 2027 for exactly this reason: organizations are automating broken processes instead of reimagining operations.
I found this in my own business. The ten hours I got back from automation didn’t come from any single tool. They came from redesigning how work moved between tools — eliminating the manual connections, the copy-paste bridges, the “check three dashboards then make a decision” routines that no AI tool was designed to fix.
Where This Gets Practical
My content creation client eventually got it right. We ignored the tools entirely for two weeks, mapped every step from content idea to published post, and found four handoff points that existed for no reason other than habit. Removing them cut their total content pipeline from ten hours to three — without changing a single tool in their stack.
Most businesses I work with have a version of this — an eight-hour bottleneck hiding behind a two-hour task. It’s normalized friction. It’s the process that “works fine” until you actually measure it.
If you’re not sure whether your AI tools are underperforming or your workflows are, that’s the work I do. No pitches, just a conversation about where the friction actually lives.
Want more like this? I write about AI adoption for people who run things — not hype, not doom, just what’s actually working. Subscribe here.
Sources
- PwC, “2026 AI Business Predictions,” 2026.
- Deloitte, “State of AI in the Enterprise,” 7th edition, January 2026.
- Deloitte, “Tech Trends 2026,” via Digital CxO, December 2025.
- PwC, “29th Annual Global CEO Survey,” via World Economic Forum, January 2026.
- Deloitte, “Humans × Machines: Work Design Is Essential to AI ROI,” October 2025.
- HR Executive, “Scaling AI in SMBs: Measurable Gains and Predictions for 2026,” December 2025.
