The AI-to-P&L Note · No. 1

The AI value gap is an operating-model gap

Boston Consulting Group’s 2025 research found that 60% of companies report little or no value from AI, while only 5% achieve value at scale. The difference is rarely the model. It is how the work is organised around it.

Watch the 36-second version

What the numbers say

BCG’s The Widening AI Value Gap (2025) is blunt. Six in ten companies say they get little or no material value from their AI investment. One in twenty is getting value at scale. The rest sit somewhere in between, running pilots and waiting for the return.

That distribution is not explained by access to technology. Most large firms, and almost every South African bank and insurer, can buy the same models, run them on the same cloud platforms and hire from the same integrators. If access were the advantage, the gap would close as the technology commoditises. It is widening instead.

When everyone can buy the same model, the model cannot be the source of the difference.

Access is not advantage

What separates the small group that captures value from the large group that does not is the set of choices made around the technology: which processes to change, who owns the outcome, how decisions are re-allocated between people and systems, and how the whole thing is controlled. Those are operating-model choices. They sit with the executive team, not the data science team.

In financial services, the processes where value sits are well known: onboarding, credit decisioning, claims, collections, fraud, complaints and advice. They have volume, rules and data, which is why they attract AI. They are also heavily regulated and directly affect customers, which is why a tool added on top of an unchanged process so often stalls at the risk committee.

Where the value leaks

  • The benefit has no home. A pilot saves analyst time, but no budget line is reduced and no capacity is redeployed. The saving is real and the P&L does not move.
  • The process is unchanged. The model makes a recommendation; the old approval chain still runs in full. Cycle time barely shifts.
  • Nobody owns the number. IT owns the platform, the business owns the process, finance owns the target, and none of them owns the AI outcome.
  • Governance arrives last. Privacy, conduct and model-risk questions are raised after the demo, and the programme restarts.

Four questions that separate the 5% from the 60%

  1. Which P&L line moves, from what baseline, and who signs for it? If the answer is “productivity”, the case is not finished.
  2. Which steps in the workflow change, and what work stops? Value comes from removed work, faster decisions or better decisions, not from added tools.
  3. Who is accountable when the system is wrong? Decision rights, escalation and the human role must be designed, not assumed.
  4. What will the board and the regulator see? Controls, monitoring and the treatment of automated decisions under POPIA section 71 should be specified before build, not after.
What to do this quarter

Pick two core processes, not twenty use cases. Baseline them in money and time. Redesign the process on paper before choosing a tool. Build the governance pack in parallel. Fund the work in stages, released against measured change in the baseline.

The point

The AI value gap is an operating-model gap. Closing it is less about finding a better model and more about making the harder organisational decisions that let any reasonable model pay for itself. That is executive work, and it is where we spend our time.

Open the AI-to-P&L Calculator

Sources

  1. Boston Consulting Group, The Widening AI Value Gap, 2025: 60% of companies report little or no value from AI; 5% achieve value at scale.
  2. Protection of Personal Information Act 4 of 2013 (South Africa), section 71.
Next step

Apply this to your organisation.

The Operating Model Diagnostic gives your executive team a prioritised, governed view of where AI can move the P&L. Pricing on request.

Request an Operating Model Diagnostic