McKinsey and BCG Both Say It: Workflow Redesign Is the Difference Between AI That Works and AI That Wastes Money

Two of the world's most cited strategy firms published their 2025 AI research independently. They surveyed different companies, used different methodologies, and framed their findings differently. They reached the same conclusion.

According to McKinsey's 2025 State of AI report, the single biggest differentiator between companies capturing real value from AI and those that aren't is workflow redesign. Companies that fundamentally redesign their workflows are 2.8 times more likely to report meaningful EBIT impact. BCG found the same pattern. In their "Closing the AI Impact Gap" research, firms that move beyond deploying AI tools to actually reshaping how work gets done see efficiency gains of 30% to 50% - compared to 10% to 20% for companies that add AI onto existing processes without changing anything underneath.

The Gap Has Nothing to Do With Technology

Most businesses are treating AI as an upgrade to existing operations. Buy the tool, train the team, integrate it into what already exists. That's the deployment instinct, and it produces marginal results. McKinsey's data makes this stark: 88% of companies now use AI in some capacity, but only 6% report the kind of EBIT impact that justifies the investment. The gap between "we use AI" and "AI is changing our results" is enormous - and it has nothing to do with access to technology.

BCG calls it the impact gap. According to their analysis, AI success is roughly 70% dependent on people and processes, 20% on technology infrastructure, and 10% on the algorithms and models themselves. Most organizations are spending their attention - and their budget - in the wrong third.

In practice, this plays out in a recognizable pattern. A company adopts a generative AI tool. A few people use it to draft emails faster or summarize meeting notes. The productivity nudge is real but scattered. Nobody has mapped which workflows it should change, how outputs should be validated, or who owns the process once the tool is running. Twelve months in, the tool is still being used. The business results haven't moved.

Operations leaders who have seen this before recognize it immediately. The tools work fine in isolation. What doesn't work is bolting capable technology onto a fragile process and expecting the process to improve itself.

What the 6% Are Actually Doing

McKinsey's high performers - the 6% generating EBIT gains of 5% or more - took a different approach. They started by identifying which workflows were mission-critical, which were most dependent on manual judgment or institutional knowledge, and which had clear inputs and outputs that AI could act on. Then they redesigned those workflows around what AI can actually do, rather than fitting AI into what already existed. 55% of high performers report having done this kind of fundamental redesign. Only 20% of everyone else has.

The workflow redesign mindset starts with a different question. Not "where can AI help?" but "which processes, if redesigned, would have the biggest impact on our operation?" That framing shift matters because it moves the conversation from features to outcomes.

In practice, a well-scoped redesign looks something like this. A process that one person has managed manually for years - order entry from emailed purchase orders, monthly commission reconciliation that takes two days of spreadsheet work - gets mapped end-to-end. What are the inputs? What decisions get made? Where does the output go, and who acts on it? Once the process is visible, it becomes clear what AI can handle reliably, what still needs human review, and what needs to be rebuilt before automation makes sense.

BCG's research shows that companies doing this well spend more time on training and defining human-in-the-loop checkpoints than on tool selection. McKinsey's data agrees: high performers are nearly three times as likely to have defined validation processes for AI outputs. The implementation comes later. The redesign comes first.

This doesn't have to be a slow or expensive process when it's scoped correctly. Linea typically works with clients to identify one or two high-value workflows in the first engagement - not a company-wide overhaul, but a specific process where the inputs are clear, the stakes are defined, and the ROI is measurable. That first workflow builds the organizational muscle for the next one. Companies that struggle are the ones that try to redesign everything at once, or the ones that skip redesign entirely and wonder why their tools aren't delivering.

BCG flagged something important on measurement: 60% of companies lack defined financial KPIs for their AI initiatives. Without those, it's impossible to know whether a workflow is performing better or just differently. Before any redesign goes live, the measurement framework needs to be in place - what gets tracked, who reviews it, and what threshold triggers a process review.

The Convergence Worth Paying Attention To

McKinsey and BCG aren't software vendors with tools to sell. They're surveying thousands of business leaders across industries and geographies, looking for patterns in what's actually working. The pattern is consistent. Companies getting real results from AI aren't necessarily the ones with the biggest budgets or the most sophisticated models. They're the ones that decided - before anything else - which processes needed to change, and then changed them.

For operations leaders at mid-sized companies, that's both a challenge and a clear advantage. Enterprise-scale organizations are slow to redesign workflows. They have legacy systems, political complexity, and change management overhead that makes even a focused redesign a multi-year project. A mid-sized business can identify a high-impact process this week and have a redesigned workflow running in production in weeks - not quarters. Speed and simplicity are structural advantages. The question is whether your organization is using them.

Frequently Asked Questions

What does workflow redesign actually mean for a mid-sized business?

Workflow redesign means mapping an existing process end-to-end - its inputs, decisions, outputs, and the people involved - and then rebuilding it around what AI can reliably do, rather than adding AI tools on top of how the process already works. For a mid-sized business, this might mean rethinking how purchase orders get processed, how job costing reports get compiled, or how customer inquiries get routed. The goal is a process that's faster, less dependent on any single person, and produces outputs that can be tracked against defined KPIs.

Why do so many AI deployments fail to deliver ROI?

According to BCG's research, 60% of companies generate no material value from AI investments despite adopting the tools. The most common reason is that organizations deploy AI onto existing processes without redesigning those processes. The tools function, but the workflows underneath are fragile, undocumented, or reliant on institutional knowledge the AI can't access. McKinsey found that only 6% of companies report meaningful EBIT impact - and those companies are 2.8 times more likely to have fundamentally redesigned their workflows before or during implementation.

How should an SME decide which workflows to redesign first?

Start with processes that are high-volume, currently manual, and dependent on one or two specific people. Those three characteristics signal both the most risk and the most opportunity. A process that runs the same way hundreds of times a month, relies heavily on someone manually handling inputs, and slows or breaks when that person is unavailable is a strong first candidate. The first redesign doesn't need to be the most complex - it needs to be specific enough that the ROI is measurable and the risk is contained.

Is workflow redesign something a business can do internally?

Some businesses have the internal capacity to map and redesign workflows, particularly if they have operations or continuous improvement experience. The challenge is that internal teams are often too close to existing processes to see where the real redesign opportunities are. They also rarely have direct experience with what commercial-grade AI can and can't do reliably in a production environment. An implementation partner who has deployed similar workflows elsewhere can compress the learning curve significantly and help avoid the most common failure modes.

What's the practical difference between deploying AI tools and redesigning workflows?

Deploying tools means adding AI capabilities to a process that otherwise stays the same - a team uses an AI tool to draft content faster, or a generative layer gets added to an existing interface. Workflow redesign means rethinking what the process is, how it runs, who's involved, and what success looks like - then building AI into that new process from the start. According to BCG, tool deployment typically yields 10% to 20% productivity improvement. Workflow redesign yields 30% to 50%. The gap is the difference between augmenting an old process and building a better one.

Book a strategy session with our team at bylinea.com to start with the workflows that matter most to your business.