Why data products must start with decisions.

Data products are not best designed from the data that happens to be available. They are best designed from the decisions the organization needs to make better, faster, more consistently, or more safely. The decision defines what the data product must contain, who must own it, how it must be governed, and how trust must be visible.

Organizations that invert this — starting from data and hoping decisions follow — typically end up with more dashboards, more pipelines, more cost, and the same uncertainty in the executive committee.

"Data-driven" describes activity — more data, more reports, more analysis. "Decision-driven" describes intent — better decisions, made by named people, supported by trusted data products designed for that purpose. The first is measured in outputs. The second is measured in outcomes.

When AI enters the picture, the difference becomes critical. AI agents and copilots do not just consume data; they act on it. They need data products designed around the decision they are participating in, not generic feeds of whatever was easiest to extract.

• MythStart from the data you have.

  RealityStart from the decisions you need to improve. Available data is rarely the same as relevant data.

• MythMore data means better decisions.

  RealityDecision quality comes from trusted, contextual data products aligned to a specific decision — not data volume.

• MythDecisions are a downstream concern.

  RealityDecisions are the design input. The decision defines what the data product must contain, govern and explain.

Data Tiles takes a decision-driven position. Every trusted data product should be traceable to a named decision, a named decision-maker, and a named business outcome. That is how data products earn their cost, their priority and their permission to scale.

Decision-driven design also creates the right conditions for AI. When the decision is explicit, the data product can be governed around that decision, and AI agents using it have a clear scope, clear accountability and clear boundaries.

1. 1

   Name the decision first

   Write the decision in plain language before any data work begins — who decides, how often, with what risk and what outcome.

2. 2

   Identify the decision owner

   The accountable decision-maker is also the natural business owner of the resulting data product.

3. 3

   Define decision context

   Capture the policies, exceptions, customer realities and operational pressures that shape how the decision is actually made.

4. 4

   Trace data back from the decision

   Identify the smallest, most trusted set of data needed to support the decision — not everything that exists about the topic.

5. 5

   Design for reuse across related decisions

   Group related decisions into a domain so the data product compounds in value rather than being a one-off.

6. 6

   Measure decision impact, not data volume

   Track improvements in speed, confidence, consistency and outcomes — not rows produced or dashboards built.

• What decision will this data product change?
• Who owns that decision today, and how is it made now?
• What does "better" look like — faster, more consistent, less risky, more explainable?
• How will we know the data product improved the decision?
• Could this data product support related decisions in the same domain?

Latttice helps business teams create and govern the trusted data products their decisions depend on. Lenz extends that foundation into governed AI agents and decision-supporting workflows. Together they operationalize decision-driven thinking — turning it from a slogan into a working operating model.