The Art of Decision Making: what analytics will mean in 2036

What will analytics look like in ten years? It sounds like a manageable question. When Anouk Gorris, our VP of Products, sat down with a coworker to answer it, they got stuck on something more basic: every time one of them said "analytics", the other heard something slightly different. A dashboard. A data team. A discipline. A market category. A line item in a budget.

Anouk trained as a linguist before she worked in data, and that disagreement became the starting point for The Art of Decision Making: Decision-Making Yesterday, Today and Tomorrow, a new research paper from Datalogz Research.

The long way round

To forecast what analytics will mean in 2036, the paper first asks what it has meant so far. It traces the word from Aristotle's Prior and Posterior Analytics, through clay tokens at Uruk, the Domesday Book and Pacioli's double-entry ledger, to Florence Nightingale's polar-area diagrams and the analytics market of 2016. It then sets that history against what Gartner, Forrester and practitioners on Reddit and Hacker News are saying in 2026.

What the research found

The name keeps changing. Since 1958 the activity has been relabeled roughly once a decade: business intelligence, data warehousing, data mining, analytics, big data, self-service, the semantic layer, and now agentic analytics and decision intelligence. Each new label tracked a change in who was buying the tools.

The problems stay the same. In 2016, Gartner expected fewer than 10% of self-service BI initiatives to be governed well enough to prevent inconsistencies. In 2026, practitioners still describe executives receiving two different churn figures in the same board meeting. Ten years of new technology produced the same list of complaints, and most of them come down to disagreement about what the numbers mean.

AI raised the stakes. A dbt Labs benchmark from April 2026 puts text-to-SQL at 84 to 90% accuracy, against 98 to 100% when queries run through a semantic layer. The difference in how each one fails matters more: text-to-SQL returns a plausible but incorrect answer, while the semantic layer returns an error message. A copilot answering confidently from a metric nobody agreed on is a harder problem to catch than a dashboard nobody opens.

What changes by 2036

Anouk expects three shifts. Agents will start acting as both interpreter and decider, mostly for low-stakes, high-frequency decisions that nobody audits later. The semantic layer will become contested infrastructure, since every platform vendor now ships its own. And consumption pricing will turn cost into a first-class governance concern, because ungoverned analytics now arrives with a monthly invoice.

The paper closes with three scenarios for 2036: the platform wins, the semantic layer wins, or the spreadsheet wins again. Its own bet is an uneasy mixture, where each platform keeps its own layer of meaning and an independent layer reconciles them.

Where Datalogz fits

The paper is open about where its author stands. Datalogz sits above data platforms, BI tools and semantic layers, at the point where a number becomes a decision. Our job is to make every decision traceable, consistent and defensible after the fact, whether it comes from a dashboard, a deck, a spreadsheet, a person or an AI agent.

Read the full paper

The Art of Decision Making runs to 18 pages plus 62 cited sources, covering 5,000 years of decision-making, the 2026 analyst landscape and a forecast for the decade ahead. If you lead a data, analytics or AI function, it is a useful frame for the conversation your team is probably already having about semantics, agents and ownership.