Companies worldwide are pouring billions into Artificial Intelligence, but a critical question looms: Is the investment paying off? Many businesses track AI adoption, but this only measures activity, not impact. A new approach is needed.
The Flaw in Counting Clicks
For years, the
success of a new technology was often measured by adoption. How many employees are using the tool? How many teams have access? These are known as 'vanity metrics'. They look good on a presentation slide but fail to answer the most important question: Is the business actually better off? A recent McKinsey report highlights a growing gap between AI adoption and measurable economic returns, noting many organizations focus on implementing technology rather than assessing the value it creates. This is a critical distinction because high usage doesn't automatically translate to improved revenue, lower costs, or a better customer experience. An employee can use an AI tool all day, but if the work process itself remains inefficient, the true value is lost.
From Activity to Achievement
In response, consulting giant McKinsey is championing a shift in perspective: from measuring activity to measuring achievement. Instead of just asking if AI is being used, the focus is now on what that usage accomplishes. This approach connects AI deployment directly to business outcomes. According to McKinsey, this requires a framework that establishes a clear, auditable line from the technical performance of an AI model all the way to its financial impact. The new methodology, detailed in reports like 'The AI Value Curve', compares the cost of performing a task with human labour versus the full cost of doing it with AI, including necessary human oversight and rework. This reveals where AI is most cost-effective and provides a more honest assessment of its economic benefit.
What Outcome-Based Measurement Looks Like
In practice, this means moving beyond simple counts. For example, instead of tracking the 'number of customer service chats handled by a bot', an outcome-focused metric would be 'the percentage increase in first-contact resolution' or 'the reduction in customer wait times'. McKinsey outlines a five-layer framework to achieve this, starting from the bottom up. It begins with technical performance (like latency and accuracy), moves to user adoption, then operational KPIs (like cycle times), strategic outcomes (like customer retention), and finally, financial impact (like revenue uplift and cost reduction). This structured approach ensures that every layer of AI implementation is tied to a tangible business goal, making it clear whether the technology is just active or truly effective.
Why This Changes AI Strategy
This shift from usage to outcomes has profound implications for how companies should approach their AI strategy. It forces leaders to define what they want to achieve before they even deploy a tool. McKinsey’s research suggests that a 30% increase in AI costs could erase $850 billion in potential annual global value, underscoring the importance of deploying AI not just widely, but wisely. The analysis also reveals that AI’s economic benefits are not uniform; they are local. A solution that is cost-effective in a high-wage economy may not be in a market with lower labour costs. This nuanced view helps companies avoid costly mistakes and focus their AI investments where they will generate the highest return. It also highlights that some of the biggest gains may come from supporting highly skilled professionals, not just automating lower-wage tasks.
