The 5% Problem
BCG's Widening Gap study buried the most important number in the executive summary: 5% of companies are generating meaningful EBIT impact from AI. Not 40%, not 25%. Five percent. The question isn't why AI isn't working. It's what those five percent are doing differently.
- AI strategy
- BCG
- business transformation
- AI value
BCG published its "Widening Gap" study in September 2025 and buried the most important number in the executive summary: 5% of organizations have captured substantial, measurable financial gains from AI. The remaining 95% have access to the same models, the same cloud infrastructure, and in many cases the same vendors. If the gap were about technology access, it would have closed by now. It hasn't. The gap is widening.
BCG identifies four attributes that distinguish AI leaders from the rest: governance maturity, data quality, talent investment, and executive commitment. The attributes are real. The problem is that naming attributes of successful organizations isn't the same as describing how they got there. You can read the study, recognize your own deficits in each dimension, nod through a board presentation, and still not know what to do on Monday. The gap between "we score poorly on governance maturity" and "here is the first thing to change" isn't a gap BCG closes.
What the 5% did first
There's a prior question the study doesn't ask: what did the 5% do first?
The attribute description tells you what they look like now. It doesn't tell you the sequence that produced the attributes. Governance maturity, data quality, talent investment — these are downstream of decisions, not the decisions themselves. The useful question is what the first decision was, and whether it was different in kind from what the 95% decided.
Looking at the case work visible in the BCG data, the pattern is consistent. The 5% didn't deploy enterprise AI licenses and then figure out where to use them. They identified a workflow that mattered, mattered in the specific sense that it was high-volume, manual-intensive, and central to how they made money or served customers. Then they redesigned that workflow around AI execution, completely, before they scaled to anything else.
The 95% did it the other way around. They acquired the capability, then went looking for places to apply it. The incentive structure of enterprise software sales is designed to produce exactly this sequence: here's a platform, now decide what you want to do with it. When you buy the capability first and ask questions later, you end up with AI tools sitting alongside existing processes, occasionally consulted, rarely embedded, and almost never in a position to restructure how the work actually gets done.
The prescription BCG states in passing
BCG's own report contains the prescription in almost parenthetical form: "start with high-impact use cases." It's stated as if it's obvious. It isn't. The reason it gets ignored in favor of broad deployment is that broad deployment looks like progress from the outside, fits the story a software vendor needs to tell to justify an enterprise contract, and doesn't require the difficult internal work of actually deciding which workflow to change first.
Deciding which workflow goes first is harder than it sounds. It requires someone who understands both the AI capability and the specific business well enough to identify where the hours are buried, where the manual density is highest, where a redesigned process would produce the cleanest data output and the most defensible ROI. That judgment isn't in a BCG report. It's not in a readiness assessment. It's in a discovery engagement with an agency that has done it before, for a company with a similar profile, and can show the before and after.
This is what BCG's research implies but doesn't name. The path from 95% to 5% goes through a single redesigned workflow, not a company-wide transformation initiative. The company-wide initiative is what the 5% looks like at scale. It's not the entry point. The entry point is smaller, more specific, and more dependent on having the right agency than on having the right budget.
The scale-first mistake
The scale-first mistake has a seductive logic. A CEO who has decided AI matters wants to move quickly. The most visible action available is a large procurement: enterprise licenses, a big-name vendor, a transformation program with a steering committee. This looks like serious commitment. It often produces, eighteen months later, a careful accounting of tools deployed, usage rates that plateau at 20–30% of licensed users, and a question nobody wants to answer in the board meeting about where the value is.
The companies in BCG's 5% look, in retrospect, like they had a strategy. They did. But the strategy was less about foresight than about starting from the right place. They asked "what's the highest-value thing we could change about how this work gets done?" and then changed it. The lesson isn't strategic sophistication. It's decision sequence.
The 95% still can close the gap. But the path runs through a different first bet. Not "which tools should we buy" but "which workflow should we redesign, and who has done that redesign before for a company like ours." The BCG data is clear on the size of the gap. The entry point it implies is a single well-scoped discovery engagement, not a transformation program.
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.
