Access to tools is becoming widespread. The ability to scale their use remains very uneven.

A minority captures a disproportionate share of growth

In the categories analysed by Bain in 2025, 120 insurgency brands captured 39% of the growth of their categories, holding about 2% of the market share.

Bath · Insurgent Brands · 2025 · categories analysed by Bain — View source

These results show that the initial size alone does not guarantee the capture of growth: relatively small players can capture a disproportionate share of the growth in their category. They do not say why, and it would be imprudent to have them say that these actors are gaining because of their technological maturity.

Adoption of AI evolves at two speeds

The second phenomenon is distinct from the former, even if it converges with it. In a study on the maturity of enterprises — all areas combined, not just marketing — McKinsey notes that 92% plan to increase their investment in AI, but only 1% actually consider themselves mature.

McKinsey · Superagency in the workplace · 2025 · companies surveyed, maturity AI all areas — View source

The same gap is seen within a single sector, which makes it even more demonstrative. BCG and the Consumer Goods Forum note that 75% of consumer goods players remain in exploration or piloting, with only 18% actually going to scale; among retailers, 45% are already in scale phase while 40% have barely started.

BCG and Consumer Goods Forum · How CPG and retail leaders maximize AI ROI · 2026 · CPG actors studied — View source

BCG also ranks only 5% of the enterprises surveyed as future construction, while 60% still derive little material value from AI. Leaders gain up to 5× more revenue gains and 3× more cost savings — a high, not an average.

BCG · Are you generating value from AI? The wide gap · 2025 · enterprises studied — 'up', high values — View source

The barrier is the operational capability

It is the heart of the demonstration. Access to tools is becoming commonplace; what remains uneven is the ability to simultaneously gather data, AI, talent, technology, external expertise, processes and governance.

• Resources

  Data exploitable, technology, models, budgets.

• Skills

  Internal talent, agencies, specialists, contractors.

• Organization

  Process, responsibilities, governance, learning capacity.

BCG summarizes this distribution by a 10/20/70 framework, used to emphasize that the majority of the effort and value of an AI transformation depends on people, processes and organization — 10% of algorithms and 20% of technology. It is a reading framework, not a universal empirical decomposition.

BCG · AI transformation is a workforce transformation · 2026 · BCG analytical framework — View source

This is true on the tool side: 34% of the Martech decision-makers in large companies interviewed cite the lack of skills as a major obstacle. Buying a stack does not create an operational capability.

McKinsey · Rewiring martech: from cost center to growth engine · 2025 · Martech decision-makers from large companies surveyed — View source

Finally, the OECD documents a persistent gap in AI adoption between SMEs and large enterprises.

OECD · AI adoption by small and medium-sized enterprises · 2025 · comparison SMEs / large enterprises — View source

Another OECD figure deserves to be read separately: 42% of the SMEs surveyed identify digital marketing and SEO as their most urgent need for digital training. It documents a digital skills gap, not directly an AI maturity gap.

OECD · SME digitalisation for competitiveness · 2025 · SMEs surveyed — lack of digital skills — View source

Access to an ecosystem of expertise, or isolation

McKinsey describes modern marketing as a hybrid model that combines in-house talent, agencies, specialists, and contractors.

McKinsey · Modern marketing: what it is, what it isn't · 2025 — View source

A sophisticated marketing organization now assembles internal teams, data, martech, AI, agencies, specialists and governance. Those who know how to orchestrate this ecosystem have a structural advantage. Those who cannot reconstruct it must find another way to access this sophistication.