Metrics, Insights, and AI: A Decade of Data Transformation
Metrics, Insights, and AI: A Decade of Data Transformation
In 2012, Irfan Kamal’s HBR article “Metrics Are Easy; Insights Are Hard” highlighted a crucial challenge in the data-driven business world.
Back then, we used to say that finding actionable insights is like finding needle in a hay stack as there was so much of data. The tools and techniques that were available back at that time did not make it easy for teams to glean insights from data.
Fast forward to 2025, and while the core message still resonates, the landscape has dramatically evolved. Today, we arre witnessing a paradigm shift. Large Language Models (LLMs) and AI have revolutionized how we process data and generate insights. These advanced systems can now interpret complex datasets, offering nuanced analysis and actionable recommendations through natural language queries. This relates to both predictive and prescriptive capabilities.
Kamal’s examples of the Netflix recommendation engine and low clickthrough rates for online ads seem almost quaint now. In 2025, purpose-built LLMs are tailored for specific industries, extracting data from documents with unprecedented precision. For instance, in the lending industry, these models can interpret loan agreements and highlight compliance risks in minutes, a task that once required extensive human effort.
In the automotive industry, I have seen first hand, how the predective analytics can reduce automotive manufacturer’s operational costs and risk in finding safety issues sooner while imrproving safety and compliance.
Kamal’s article provides a four-step process (collect, connect, manage, analyze) and remains relevant, but it has been supercharged by AI advancements. AI now gathers data from diverse sources, including multimodal inputs like text, audio, and images. Advanced AI systems link data to specific segments while ensuring compliance with evolving regulations. AI-driven Enterprise Content Management (ECM) systems now offer automated content summarization and intelligent metadata tagging.
Generative AI and specialized LLMs collaborate with human teams, uncovering insights that were previously unattainable.
“You can have all of the fancy tools, but if [your] data quality is not good, you’re nowhere.” - Veda Bawo, director of data governance, Raymond James
The landscape of actionable insights has transformed dramatically since Kamal’s original observations. In 2025, AI-driven personalization has reached new heights, particularly on platforms like TikTok and Instagram. For instance, TikTok’s AI doesn’t merely suggest content based on friends’ interests; it creates hyper-personalized video feeds in real-time, adapting to users’ micro-interactions and viewing patterns1. This level of personalization has led to an average of 95 minutes spent daily on the app, showcasing its effectiveness in engaging users.
The tools and techniques are readily available from cloud providers such as AWS and Microsoft Azure. They offer powerful personalization services that leverage AI and machine learning to enhance user experiences.
AWS offers Amazon Personalize, a machine learning service that generates personalized recommendations for applications. It can be used to create individualized product recommendations, tailored search results, and customized direct marketing. Amazon Personalize can be integrated with other AWS services and doesn’t require extensive machine learning knowledge to implement.
In this new era, the challenge isn’t just finding insights. It’s finding ways to harness and use AI-generated actionable intelligence. The focus has shifted from struggling to find insights to effectively implementing the wealth of insights now available at our fingertips. As we navigate this data-rich landscape, the key to success lies in our ability to harness these AI-driven insights and translate them into strategic action.
In this new landscape, organizations have a unique opportunity to establish robust processes for gaining insights and implementing strategic actions aligned with their business objectives. This transformation requires leadership recognition and the allocation of appropriate resources. To fully leverage the advances in AI, companies must invest in the right talent and develop complementary processes. By doing so, they can effectively manage this change and harness the full potential of AI-driven insights for strategic decision-making.
- Metrics Are Easy; Insight Is Hard (https://hbr.org/2012/09/metrics-are-easy-insights-are-hard)
- Data and Intuition: Good Decisions Need Both (https://www.harvardbusiness.org/data-and-intuition-good-decisions-need-both/)
- Data is only Valuable if Insights informs (https://www.linkedin.com/pulse/data-only-valuable-insights-informs-jay-milligan-1e/)
- AI Personalization Examples That Will Surprise You (https://www.idomoo.com/blog/ai-personalization-examples-that-will-surprise-you/)