The 25-Point Spread
BCG found that strategy plus workflow redesign lifts measurable business impact by 25 percentage points above baseline. Better tools without redesign: 5 points. The 20-point gap between them is what implementation actually is.
- BCG
- workflow redesign
- AI implementation
- ROI
- operating model
BCG published a finding in June 2026 that's worth stopping on. Organizations that combine a clear AI strategy with workflow redesign see measurable business impact 25 percentage points higher than baseline. Organizations that deploy better AI tools without workflow redesign: 5 points. The gap between those two numbers, 20 points, is what implementation actually is.
The 5-point result isn't nothing. Tools without redesign do produce some improvement. The error is treating it as the destination.
A tool versus an operating model change
The 5-point result is what you get when you add AI to an existing process. The tool helps people do what they were already doing, faster. They produce more drafts, check more sources, write better first cuts. The process is the same. The handoffs are the same. The decision points are the same. The data output is the same. The improvement accrues to individuals in the form of time savings, which may or may not show up in organizational output.
The 25-point result is what you get when the process changes. The workflow is redesigned around AI execution: which steps the AI runs autonomously, which require human review, what the handoff protocol looks like, what data gets created at each stage. The improvement accrues to the process, not just to the individuals running it, which means it compounds and scales in ways that individual productivity gains don't.
This isn't a subtle distinction. It's the difference between a tool and an operating model change.
What redesigning the workflow actually requires
The practical question is what "redesigning the workflow" actually requires before you start. Most implementation conversations skip it. Here's what it isn't: picking a platform, scoping the automation, and writing prompt instructions. Those are outputs of redesign work, not the redesign itself.
The redesign work starts with the current workflow as it actually runs — not the description on the org chart, but the thing that happens when a real input arrives on a real Monday morning. What is the decision point? Who owns it? What data does the decision require, and where does that data currently live? What does "wrong" look like, and what happens when the AI is wrong? Who catches it?
A 200-person distribution company that automated its invoice exception handling in 2025 got the 5-point result first. The tool caught exceptions and flagged them for human review. It was faster than before, but the review queue was the same size. The redesign happened six months later: the workflow was rebuilt around the categories of exceptions the AI was reliably handling, reassigning human review to genuinely ambiguous cases and closing the loop on routine ones automatically. Same tool. The 20-point gap showed up in the second phase.
Scope around the workflow, not the tool
The BCG data implies a sequencing question that most implementation briefs never ask: are you scoping around the tool or around the workflow?
Scoping around the tool starts with the platform ('we want to implement Agentforce for lead routing') and works backward to the workflow. The scope is defined by the tool's capabilities. The workflow gets fitted to the tool's parameters.
Scoping around the workflow starts with the decision: "We want leads routed to the right rep within four minutes of submission, with no human intervention for standard profile matches." The tool gets selected after the workflow requirement is defined. The scope is defined by what the process needs.
The first approach gets you an implementation. The second gets you a redesign. BCG's data suggests the difference is 20 percentage points of measurable impact.
The deliverable that tells you which one you're buying
An agency running a real redesign engagement produces a specific deliverable before the build starts: a current-state workflow map with every decision point, data input, and handoff identified, alongside a future-state design that specifies what the AI runs, what the human reviews, and what the protocol looks like for exceptions.
If the agency goes directly from a kickoff call to a build scope without that map, they're implementing around the tool, not redesigning around the workflow. BCG's 25-point result won't follow.
The map is the signal. Ask to see a redacted example from a prior engagement before you sign.
If you want that question answered for your specific situation, the Forge Playbook does it. Answer a few questions about your business and we'll put together a tailored outline of which workflows are worth automating and what a realistic budget looks like for each. Free, no obligation, takes about three minutes.
