How Kenyan businesses can turn tech into measurable gain

Businesses should focus not on whether to adopt AI, but on where it can deliver measurable economic value.[Courtesy]

The most important question facing Kenyan businesses in the current wave of artificial intelligence (AI) adoption is not whether to invest in the technology, but where that investment will create measurable economic value.

Too much of the current conversation starts with the technology and works backwards to find a problem it can solve. When companies see competitors investing in AI, the instinct is to follow: acquire tools, launch pilots, and celebrate adoption rates, often before the commercial problem the investment is meant to solve has been defined. That sequence should be reversed. Sound financial management starts with identifying a costly, repetitive or strategically important process, establishing how it currently performs, and only then asking whether technology can improve its speed, accuracy or financial outcome.

This distinction matters because efficiency is not the same as value creation. An automated reporting process may save employees several hours a month, but that alone does not prove the business is better off.

The more consequential question is whether faster access to information led to a better pricing decision, surfaced an unfavourable cost trend, improved capital allocation, or helped avoid an investment whose expected return did not justify its risk.

Technology is reshaping what businesses should expect from their finance teams. Recording transactions, reconciling accounts and explaining historical results remain essential, but finance professionals must increasingly help leaders understand what could happen next.

That requires combining financial and operational data, building forecasts, and assessing how alternative decisions might affect the business before a product launches or a new market is entered. Scenario and sensitivity analysis, for instance, can show how changes in price, customer adoption, costs or market conditions would affect expected returns.

Rather than presenting a single forecast as a prediction, finance professionals can model base, upside, and downside cases. This does not eliminate uncertainty; it makes uncertainty visible, and helps decision-makers identify which assumptions carry the greatest influence over an investment’s success.

In my experience supporting financial analysis in the technology sector, the value of a model lies not in the number it produces, but in how well it helps managers understand the trade-offs between growth, investment, risk and profitability before resources are committed.

Finance leaders should evaluate AI with the same discipline they apply to any other capital investment. The assessment must extend beyond the purchase price to include integration, data preparation, employee training, cybersecurity, governance, maintenance, and the financial consequences of incorrect outputs.

Success criteria should be defined before implementation, not after. Depending on the application, appropriate measures might include improved forecast accuracy, lower processing costs, stronger working-capital management, incremental contribution margin, or return on invested capital.

Without a baseline and predetermined criteria, a company may know employees are using an AI tool without ever being able to determine whether the investment improved performance. And there is a bigger risk as well. Automating a poorly designed process does not make a business smarter. It simply executes the wrong process faster and at greater scale. AI operating on incomplete or inconsistent data can produce sophisticated-looking outputs without improving the underlying decision.

In February 2024, Klarna, the global fintech company known for buy-now-pay-later (BNPL) services, reported that its AI assistant had handled 2.3 million customer conversations in its first month, roughly two-thirds of all customer-service chats. The company said the system cut average resolution time from eleven minutes to under two, and estimated it contributed approximately $40 million (Sh5.18 billion) in profit improvement in 2024.

On the surface, these figures demonstrate that automation can deliver substantial operational and financial gains. However, Klarna later renewed its emphasis on human customer-service agents, recognising that some customer needs still require empathy, context and judgment that automated interactions cannot adequately provide.

The broader lesson is that the strongest operating model is rarely full automation. It is typically a tiered system in which AI handles repetitive, high-volume activity while employees manage exceptions, complex needs and judgment calls.

Evaluating AI’s financial impact, therefore, requires looking beyond labour savings to service quality, human-intervention rates, the cost of errors, and the long-term effect on customer trust. Kenyan businesses should not simply replicate whatever global technology companies adopt.

They should weigh their own resources, operating environment, customer base, and data maturity.

The most sensible starting point is processes that are repetitive, financially meaningful, and supported by reliable data, demand forecasting, fraud detection, working-capital management, transaction reconciliation, customer service, pricing analysis and financial scenario modelling are all strong candidates.

Before investing, management should be able to answer five questions: What problem are they solving? What is the current performance baseline? How will the technology improve the process or decision? What are the full costs and risks? In addition, how will projected benefits be measured against actual results?

This discipline matters most for small and medium-sized businesses, which cannot afford to invest in technology simply because it is fashionable. For many, the right first step is modest: automating a reconciliation, building a reliable performance dashboard, or creating a structured demand forecast.

As automation absorbs more routine work, finance professionals will need to combine technical financial knowledge with data analysis, commercial judgment and clear communication. Human judgment will matter more, not less.

Competitive advantage will not automatically belong to the businesses that adopt AI first. It will belong to those that identify the right decisions to improve, build the data and controls to support them, and convert greater efficiency into better judgment and sustainable growth.

- The writer is a Finance Manager at Amazon, Seattle, USA