← Glossary

The AI-competence gap

AI tools are cheap and everywhere; the competence to use them well, and the organisational capability to turn that competence into results, is rare.

Companies are buying AI for their commercial teams and getting almost nothing back, and most cannot work out why. Adoption is close to universal, yet only a small minority of firms can trace any effect on earnings, because the tools lifted the odd individual’s output and never became an enterprise result (MIT, 2025; McKinsey & Company, 2026). The book’s diagnosis is blunt. It is not the technology. It is competence, and the book names the distance the AI-competence gap.

Where does the value actually come from?

The Boston Consulting Group puts the sharpest number on it. Three-quarters of companies have failed to scale any value from AI, and when BCG broke down what separates the winners, 10% of the difference was the algorithms, 20% the technology and data, and 70% people and process (Boston Consulting Group, 2024). The money, meanwhile, goes the other way. The Conference Board found that while 55% of workers now use AI regularly, only one in three has had any AI training from their employer in the last six months (The Conference Board, 2026). People are handed something powerful and left to work it out alone, with no method and no coaching.

Figure

Where the value from AI is won or lost.

AlgorithmsTechnology and dataPeople and process
The toolsCheap and everywhere

55% of workers use AI regularly (The Conference Board, 2026).

The competenceRare

Only one in three has had any AI training from their employer in the last six months (The Conference Board, 2026).

The capabilityRarer still

Three-quarters of companies have failed to scale any value from AI (Boston Consulting Group, 2024).

The bar splits what separates companies that scale value from AI from those that do not, as reported by the Boston Consulting Group (2024). The figures in the three panels are as cited in chapter 10 of the book; they come from different surveys and are not shares of one population.

What does the gap look like on a team?

An illustration. A commercial team of forty is given licences on a Monday. Within a month six people are producing pre-call briefings that would have taken a day, built in the customer’s own numbers. Most of the rest use the tool to tidy emails. Nobody has written down how the six do it, the account plans are not connected to anything, and the quarter’s results look exactly like last quarter’s. Individual output went up. The enterprise result did not move.

Both halves of the gap are visible there. Competence belongs to each person: the skill to point the tool at a customer problem and judge what comes back, the territory of AI fluency. Capability belongs to the organisation: a shared method, data the tools can reach, and work redesigned so the six become the norm rather than the exception. Those are the two axes of the hidden edge, and the book says outright that the matrix was a picture about AI all along.

Where does it sit in the method?

It is the clearest argument for the customer management function. AI is not an eleventh capability on the Customer Management Navigator; it runs through all ten, from writing the strategy to modelling a price, and the book points to it capability by capability because that is the only place it earns its keep. The 70% is, in the book’s words, “a leadership number”: whether AI becomes part of how people think or another tool bolted on and ignored is decided by whether leaders build it into the work and coach people through the change. The Value Edge Diagnostic tests the same thing in its Embracing AI theme, adoption against commercial benefit.

What it is not

It is not a technology gap, so a better model will not close it. It is not a training budget either. Train the people without changing how the work is organised and the business moves into the frustrated-heroes corner: competent individuals, no repeatable result, and the best of them soon looking elsewhere.

Set out in full in: Creating Customer Value Propositions with AI (Kogan Page, 2027), chapters 2, 10 and 12.

Related terms

See the method run on one real account in three days: the One Page Proposition.