AI ROI: Why Most Companies Are Not Getting Value from AI, and How African Firms Can

October 15, 2026
AI Insights
6 min read

By GAA Editorial Team

Almost every large company now uses AI. Very few can show what it has done for the bottom line.

McKinsey’s 2026 State of AI survey of 1,719 respondents in 97 countries found that nearly nine in ten organisations use AI regularly in at least one business function. Yet only 37% reported any impact on EBIT from AI, and only about 6% qualified as “high performers”, meaning they attribute at least 5% of EBIT to AI and describe its impact as significant (McKinsey, 2026).

BCG’s 2025 study of 1,250 senior executives tells a similar story. Only 5% of companies were what BCG calls “future-built”, getting value from AI at scale. Another 35% were scaling and starting to see returns. The remaining 60% reported little or no value so far (BCG, 2025).

Accenture’s 2025 research on 2,000 companies found an even smaller group, about 8%, that it calls front-runners: companies scaling AI across the enterprise and building it into their core strategy (Accenture, 2025).

Three firms, three methods, one message. Using AI is easy. Getting a return on AI is hard, and the gap between the leaders and everyone else is growing.

Why the gap matters

The leaders are pulling away. BCG found that its future-built companies achieved 1.7 times the revenue growth of laggards, 1.6 times the EBIT margin and 3.6 times the three-year total shareholder return. They expect about twice the revenue increase and 40% greater cost reductions in the areas where they apply AI. They were also spending more than twice as much on AI as the laggards.

Accenture found something similar. Front-runners with more than $10 billion in revenue grew revenue seven percentage points faster than companies still experimenting with AI.

For African companies, the potential prize is large. McKinsey estimates that generative AI could add between $61 billion and $103 billion a year in value across the continent, with the biggest pools in retail, telecoms, packaged consumer goods, mining, heavy industry and energy, and banking (McKinsey, 2025). Capturing that value depends on doing what the leaders do.

What the leaders do differently

Across the three studies, the same practices come up again and again.

1. They redesign the work, not just add a tool

This is the clearest finding in McKinsey’s research. About three-quarters of AI high performers had fundamentally redesigned their workflows, compared with about a quarter of other companies. BCG’s future-built companies also focus on reinventing whole processes rather than making small gains in each task.

In practice, this means asking “how should this process work if AI does part of it?” rather than “where can we add AI to what we already do?” A bank that uses AI to draft credit memos but keeps the same five approval steps saves little. A bank that redesigns credit assessment around AI, with new roles, fewer handovers and clear rules for when a person must review, can change its turnaround time and cost.

2. They focus on the core of the business

BCG found that about 70% of AI’s potential value sits in core functions such as sales and marketing, manufacturing, supply chain and pricing. Only about 13% sits in IT. Yet many companies start with IT help desks and internal chatbots because they feel safe.

For an African distributor, the bigger prize is probably in route planning, demand forecasting and credit decisions for retailers. For a bank, it is in credit, collections, fraud and customer service. Start where the money is.

3. Their leaders own it

McKinsey found high performers were twice as likely as others to say their senior leaders are committed to their AI initiatives. Accenture found that C-suite sponsorship made AI success 2.4 times more likely. BCG lists a multiyear AI strategy led from the top as the first of the five practices its leaders share.

AI programmes that sit only in IT, or with one enthusiastic manager, tend to stall when they meet resistance from other functions.

4. They measure value, not activity

McKinsey’s high performers were twice as likely as others to have defined processes for measuring the impact of AI. Many companies measure the wrong things: number of licences, number of users, number of pilots. Those numbers say nothing about returns.

Measure what the business cares about: cost per transaction, cycle time, revenue per salesperson, loan default rates, customer churn. Set a baseline before you start and review it every month.

5. They invest in people

Accenture’s front-runners had talent maturity about four times higher than companies still experimenting. BCG’s research on workers found that only 36% of employees felt adequately trained in AI, and that people who received five or more hours of training were significantly more likely to use AI regularly (BCG AI at Work, 2025).

Tools are easy to buy. People who know how to use them well, and managers who know how to lead the change, are the real constraint.

Make AI Pay is our self-paced programme on getting real returns from AI. It covers how AI-powered businesses work, finding the right use cases, building the business case and taking a pilot all the way to measurable results.

Explore Make AI Pay

What this looks like in an African firm

African companies face some particular challenges in getting AI to pay:

  • Foreign exchange. Most AI tools are priced in US dollars. When the naira or cedi weakens, costs rise fast. That makes it even more important to tie every AI investment to a measurable return and to review subscriptions often.
  • Data quality. Customer and operational data is often split across systems, spreadsheets and paper. Many AI projects fail because the data is not ready. Fixing data for one priority process is better than a company-wide data project that never ends.
  • Thin management layers. In many mid-sized firms, a small team of senior managers runs everything. AI projects compete for their attention. A small number of focused projects, each with one named owner, works better than many scattered pilots.
  • Regulation. Data protection laws in Nigeria, Kenya, South Africa and elsewhere set rules on how customer data can be used. Building these rules in from the start avoids costly rework.

None of these is a reason to wait. They are reasons to be selective.

A simple playbook

Here is a practical sequence for an African management team that wants to see returns within a year:

  1. List your top ten cost and revenue drivers. Ask where AI could change each one. Pick two or three where the value is large and the data is reasonably good.
  2. Build a simple business case for each. Estimate the value, the cost, the time to results and the main risks. Be honest about the data work needed.
  3. Redesign the process. Map how the work flows today and how it should flow with AI. Decide which decisions stay with people.
  4. Run a time-bound pilot. Set a clear measure of success and a deadline, typically 8 to 12 weeks. Stop pilots that do not meet the bar.
  5. Scale what works and train the people involved. Move successful pilots into everyday operations, with training and new ways of working for the staff affected.
  6. Report results to the board. Treat AI like any other investment, with regular reporting on what it has delivered.

The bottom line

The research from McKinsey, BCG and Accenture agrees: AI pays when companies redesign work around it, focus on core business processes, put senior leaders in charge, measure real results and invest in their people. African firms can do all of these things. The ones that start now, with discipline, have a real chance to join the small group that turns AI into results.

If you are building that capability across your leadership team, the AI Leadership Bundle combines Lead with AI, Make AI Pay and Platform Power in one package at a lower combined price.

Keep reading

Sources

Learning Portal