There’s an elephant in the room when it comes to AI. Businesses are keen to talk about how deeply they’re investing and the expected impact on their organisations. But not many want to talk about the ROI from their AI investments. Is this because it’s on the low side? Is it just too soon to say? Or do they genuinely not know?
A 2025 Forbes survey found that just 1% of companies generated a 20% or more ROI from their AI investment, while 53% reported an ROI of between 1% and 5%. Similarly, a recent MIT report, titled “The GenAI Divide: State of AI in Business 2025,” reveals that while US businesses have collectively invested between $35 billion and $40 billion in AI initiatives. Almost all of them (95%) are seeing zero return on their investments or no measurable impact on profits.
This doesn’t necessarily mean they’re not seeing benefits to their AI programmes – there are non-tangible, indirect benefits which are harder to quantify. It’s more likely that they lack the frameworks to capture the impact. This presents a huge challenge for businesses looking to secure board investment and plan strategies built around AI into 2026 and beyond.
Businesses investing in their AI journey should follow a three-step process to define, measure and generate value. Although it will be different for each business, here’s a framework to keep you on track.
Step 1: redefine what value means to your business
The Forbes study found that only 5% of US businesses say they are seeing value from their AI investments. MIT Sloan Management Review and Boston Consulting Group found that although nearly 85% of executives believe AI will allow their companies to gain or sustain a competitive advantage, less than one in four (23%) have successfully measured its actual business impact.
To unlock the full potential of AI, businesses must redefine what they mean by “value” and establish metrics that meaningfully reflect AI’s impact.
- Finance and Risk Management: AI models excel at anomaly detection: a measurable reduction in fraud losses is a direct indicator of value delivered
- Network Operations: AI can predict and prevent outages, enabling intelligent traffic management; a decrease in network downtime is a tangible performance metric
- Customer Experience: evaluate improvements using metrics such as Net Promoter Score before and after AI automation; track reductions in call centre volumes, faster query resolution times, and increased conversion rates from personalised marketing campaigns
For Colt, customer experience transformation is the key driver for much of our AI adoption: driving simplicity in the ways customers connect and engage with us, automating processes and making life easier for them. AI value creation is generated by measuring productivity gains, identifying and tracking customer experience enhancements and measuring the real business impact of our newly automated systems and processes.
Step 2: ask your AI vendors how their solution will help you deliver measurable business value and what support they offer to help you define, track, and realise your ROI.
All businesses want their customers to succeed, but vendors are not always equal when it comes to helping your business drive value from their solution.
Be transparent about what you need to achieve – the clear goals you’re setting and the impact you’re expecting the solutions to have on your teams, your customers and your business. You’ll find some vendors have structured ROI models to help businesses define, track and quantify value across their AI tools, while certain platforms have integrated analytics dashboards that track productivity gains.
It’s inevitable that businesses implementing new solutions encounter unexpected barriers preventing progress: legacy infrastructure slowing down the performance of AI solutions, or the need to retrain or upskill employees, for example. According to S&P Global Market Intelligence, the proportion of companies that abandoned most of their AI initiatives has reached 42% in 2025, up from 17% in 2024. Keeping vendors updated on roadblocks gives them an opportunity to extend and fine tune their support and ensure your value creation doesn’t fall at the first hurdle.
Step 3: measure financial gains and efficiencies
Boards want proof of business impact: value creation may not be enough for CFOs facing budget restrictions. Here are some ways you can measure and present quantified, meaningful financial metrics.
- Measure a reduction in risk: quantify savings from factors such as automating fraud detection and reducing regulatory and non-compliance risk.
- Quantify reduced operational costs and productivity gains through automation: solving pain points, reducing human error or system downtime and accelerating service provision across the business.
- Quantify operational efficiency: predictive analytics reduces maintenance costs, quite literally from the ground up, from network infrastructure costs to building heating and cooling costs.
- Measure revenue growth compare conversation rates, average revenue per user and customer lifetime value after AI implementation with baseline metrics before implementation; measure financial gains from AI-based buyer intent models to drive intelligent marketing; and – this one’s close to our own heart – quantify the metrics before and after AI-driven enhancements to the customer experience from lower churn and improved retention.
- Estimate costs from business lost to competitors with more mature AI usage strategies across their organisations. This is a tough one to face up to, but still a good measurement.
- Measure and quantify gains and contributions towards net zero goals. For example,in our global HQ, Colt House in London, we’re testing a smart building solution built on a platform which uses AI and diagnostics software to optimise our building use.
To close, there’s one dependency on which your business’ successful AI implementation is built; a dependency which is fundamental to your business generating value from AI, but one that is often overlooked: engaging your people from the very earliest stages is critical. Colt’s latest research, based on responses from over 1,000 telco employees, highlights the importance of an inclusive people-first AI strategy and outlines the risks presented if employees are left behind. The research revealed that while optimism around AI’s potential is high, so are concerns about its impact on jobs. 46% of respondents believe AI training is essential to building confidence.
Those who receive AI-specific training and regular communication feel far more secure in their jobs. Confidence grows even further when training is inclusive, accounting for different learning styles, time constraints and levels of experience. Yet just 39% of employees who use AI at work have had formal training, according to the Microsoft and LinkedIn 2024 Work Trend Index.
Measure what matters
As businesses are under pressure to realise AI’s full potential, measurement is the differentiator. Without clear, consistent metrics tied to business outcomes, even the most sophisticated AI strategies risk becoming expensive experiments. Organisations that embed value measurement into every stage of their AI journey will not only justify their investments but unlock sustainable growth, innovation, and competitive advantage: because AI doesn’t pay off unless you measure what matters.
Colt Technology Services is a leading provider of digital infrastructure and IT services, dedicated to helping businesses navigate and compete in the digital world. To discover the solutions helping IT leaders ease the pressure of transformation and move forward with confidence, visit colt.net/solutions