---
title: "A data platform is not a data lake with a dashboard"
url: "https://hurukan.com/blog/a-data-platform-is-not-a-data-lake"
description: "Copying every system into one store and pointing a BI tool at it builds a bigger pile, not a platform. A platform is defined by meaning, ownership and a way back to action."
---

# A data platform is not a data lake with a dashboard

May 19, 2026·3 min read·Editorial team

An analyst seen from the side at a long desk with three monitors showing abstract blue-and-grey tables and connecting lines, printed schema diagrams pinned to a board behind, late-afternoon light

The project brief was clear: get all the company’s data into one place. Pipelines were built from the ERP, the CRM, the warehouse system and a dozen spreadsheets into a cloud data lake, loaded every night. A BI tool went on top. The launch presentation showed every source feeding one store.

Half a year later, three dashboards show three different revenue figures and the analysts spend most of their week explaining why. The operational teams still export to spreadsheets, because the lake is a day behind and cannot change anything anyway. The data is in one place. Nothing else is.

The project solved storage and transport, which were never the hard part. The hard part is meaning (what each field actually is), ownership (who is responsible for it being right), freshness that fits the decision, and a path back to the place where the work happens. A data lake is a location. A data platform is an agreement about all four.

## Where the lake with a dashboard fails

**Copies without meaning.** Tables arrive with column names that made sense to whoever designed the source system. Nobody downstream knows which of the seven status values means “shipped”, so every analyst guesses, and they guess differently.

**No owner for any fact.** When a number is wrong, the data team takes the blame. The error is almost always in the source process, and the people who could fix it never hear about it, because the lake has no idea who they are.

**One speed for everything.** A nightly batch serves a stock decision made every hour. Or, in the expensive version, everything is streamed in real time to feed a report read once a month. Freshness is set by the pipeline, not by the decision.

**Flow in one direction.** Data flows out of the operational systems into analysis. Insight flows out of analysis as a slide. The action happens back in the source system, by hand. The loop from insight to action runs through people retyping things.

**Definitions in the reporting layer.** Every dashboard computes revenue with its own formula, so the definitions drift apart one fix at a time. How to stop that is covered in [one definition of revenue](https://hurukan.com/blog/one-definition-of-revenue).

## What a data platform is

**One model of the company that every source is mapped to.** The order in the ERP and the deal in the CRM are linked to the same objects, with their relations, in one description of the business: an [ontology](https://hurukan.com/system/ontology). New sources are mapped to the model rather than dumped next to it.

**An owner for every object and every source**, named, with a written definition of what correct means. When something is wrong, it goes to that person, with evidence.

**Freshness set by the decision.** For each use, ask what decision this data serves and how old it may be when that decision is made. Some things need minutes; most need a day; a few need a month.

**Lineage.** Every number can be traced back to its sources and the transformations applied along the way, by someone who did not build the pipeline.

**A way back.** The platform can feed a decision to the place where it is carried out, as a proposed reorder, a flagged invoice or a task with an owner, through an approval rather than through a chart and an email.

The last point is why Hurukan does not keep a separate analytical copy of the business. Its [Data layer](https://hurukan.com/system/data) is designed as connectors into one data model, with imports, sync, documents and e-invoices attached to the records they belong to. Operations and analysis read the same model, so an insight can become a proposal on the same record it was drawn from. It is in build; the claim here is the design.

## A test you can run today

Pick the number your management meeting argues about most. Trace it back on paper:

-   Which system and which table it came from, and which field.
-   Which filters were applied, and who decided them.
-   Who is responsible for that number being correct, by name.
-   How old it was at the moment it was shown.

Then ask the second half of the question. When that number shows a problem, what happens next? Count how many times the information is retyped, forwarded or re-explained before something in the business actually changes.

If the trace needs more than one person to complete, or the action runs through email and memory, you have a lake with a dashboard. The usual symptom of that is reports that nobody opens, which is the subject of [the dashboard nobody opens](https://hurukan.com/blog/the-dashboard-nobody-opens). The [Hurukan overview](https://hurukan.com/) shows where the data layer sits among the rest.

-   data platform
-   data lake
-   data ownership
-   lineage

## Keep reading

-   [One identity for people and agents: who may do what](https://hurukan.com/blog/one-identity-for-people-and-agents)
-   [The live state and the record: two memories of a company operating system](https://hurukan.com/blog/the-live-state-and-the-record-two-memories-of-a-company-operating-system)