Adding a Further Perspective: The Consumption Layer

This is a strong analysis of how the methods are diffusing, and I agree with the central thesis: data and analytics is shifting from a specialist discipline to an infrastructure layer. And the clearest divide in the industry is no longer between AI adopters and non-adopters, but between organizations building integrated analytics ecosystems, enhanced by all types of AI, and those still operating project-by-project. 

I’d like to add that I don’t believe you can talk about data and analytics without including the most impactful step of the research process, the consumption layer. The section describes in depth which methods are used and how they mature, and I believe that the most significant change relating to this still lies ahead. The next natural change is a fundamental shift in how the resulting insights are delivered to and received by people. 

My thesis is that data will be consumed differently going forward. Increasingly in text and conversation form, with sharp, precise conclusions, rather than leaving the user to interpret tables and charts. Reports and dashboards will become a complement rather than the primary entry point. This follows the report’s logic: as methods become infrastructure — always on, embedded, high-volume — the bottleneck moves from “which method” to “how does a human interpret all of this quickly and correctly.” At that point, a conversational layer becomes the natural next abstraction.

I see clear progression. ‘Chatting’ with data is the new entry point — it lowers the threshold and lets anyone ask a question and get a direct answer. Graphical reports then come into play as the next step to better reveal cause and effect and the relationships within the data. And over time, we’ll gain access to more advanced visual models that help us interpret and read data in ways we don’t today.

But I want to be clear about one decisive precondition. The value of “chat with data” stands or falls on the AI models fully understanding the context, being consistent, and not creating more confusion than a clear, human-authored guideline would. A model that answers the same question differently, or that misses the business context, is more dangerous than a traditional report because it inspires confidence without earning it. The entry point must therefore be built on a foundation of contextual understanding and consistency; otherwise, we risk automating the noise rather than cutting through it.

In short: the report convincingly shows that the methods are becoming infrastructure. I’d complement that picture by arguing that the human interface is the next battleground — and that whoever solves context and consistency in the chat layer will own how data is consumed going forward.