Skip to main content

What you’re building

Conceptual modeling is the process of describing the important things a business cares about — such as customers, orders, products, and accounts — and defining how those things relate to each other, without getting into database structures or technical implementation. It gives data modeling a clear business foundation before tables, columns, and schemas are designed. That helps the resulting data model reflect how the business actually works, use consistent definitions, and stay understandable to both technical and non-technical people. Concepts and relationships can be enriched with additional meaning to form a business ontology. That might include descriptions, synonyms, attributes, and references to where the physical data is stored. The conceptual model then provides the foundation for logical, physical, and semantic data modeling. In Snap Data Studio, open the Conceptual Model tab from the project sidebar (Concepts or Conceptual Relationships). You build the model on the canvas, edit details in drawers and forms, and optionally use AI Copilot to propose changes.

Build on the canvas

Add concepts, connect them with relationships, and arrange the diagram so the model is easy to read.

Concepts

Concepts are the business things in your domain — not database tables.

Create a concept

  1. Open Concepts in the sidebar, or click Add Concept on the toolbar (or on the empty canvas).
  2. In the Concepts drawer, open Create newAdd new concept.
  3. Fill in the Create Concept form and click Save Concept.

Edit or delete a concept

  • Click a concept on the canvas, or a concept row in the Concepts drawer, to open Edit Concept.
  • Change fields and click Save Concept.
  • Click Delete Concept in the form, or press Backspace/Delete with one concept selected and confirm. Deleting a concept also removes relationships tied to it. You can delete only one concept at a time.

Concept fields

Attributes and value types

For each attribute you can set:
  • Name (required)
  • Type (required) — pick a Value Type from the catalog, or Create custom Value Type based on a built-in type
  • Description
  • Unique across instances of this Concept
  • Identifies this Concept
Built-in Value Types include Boolean, Date, DateTime, Decimal, Float, Integer, String, Time, DateTimeTz, Object, Array, and Opaque. If the concept has linked entities, you can Map attribute to a column on a linked table. The form may show Suggestions you can Accept. Accepted mappings fix columns during generation; unmapped attributes are treated as names to find in source.

Domains

Use domains to group related concepts (for example Finance or Customer). Create domains from the Concepts drawer. Assign a concept to a domain in the concept form. You cannot delete a domain that still has concepts.

Relationships

Relationships are directed connections between concepts: a source, a verb, and a target.

Create a relationship

  1. Drag from one concept node to another on the canvas, or click Add Relationship on the toolbar / Create new relationship in the Conceptual Relationships drawer.
  2. Complete the form and click Save.
You need at least one concept before you can create a relationship (including a self-referencing hierarchy).

Edit or delete a relationship

  • Click a relationship edge on the canvas, or expand a row in the Conceptual Relationships drawer and choose Edit.
  • Update the verb, type, cardinalities, or linked entities, then Save. Source and target concepts are fixed after create.
  • Delete from the form, the drawer row, or Backspace/Delete with one edge selected (confirm in Delete Relationship).

Relationship fields

Relationship types

Uniqueness

A relationship is unique by directed source → target plus verb (case-insensitive). You can have more than one relationship between the same pair if the verbs differ.

Canvas basics

The Conceptual Model canvas is where you arrange concepts and relationships.

Toolbar

Event concepts use a distinct card style on the canvas. Domain accents appear on nodes when a concept belongs to a domain.

Use the copilot

Open AI Copilot from the header while you are on the Conceptual Model tab.

Modes

You can also choose a model tier (Fast, Automatic, or Intelligent). On an empty conceptual model in Agent mode, Copilot prompts you to describe your domain so it can propose starter concepts and relationships.

Attachments and database context

Give Copilot extra context from the composer:
  • Attach file (paperclip) — attach a PDF or supported text file (size limits apply).
  • Database control — open Add tables to Copilot to include warehouse table definitions in the message context.
You can also mention saved diagrams, documents, and code files from the composer when they are available.

Review staged changes

When Agent mode edits the conceptual model:
  1. Copilot applies a preview on the canvas (nodes and edges highlight as added, modified, or removed).
  2. When ready, you see Staged changes ready with a short summary.
  3. Review the highlights, then click Keep to commit or Revert to discard.
Staged changes are not saved until you Keep. If you try to continue with unresolved changes, Copilot asks you to Keep or Revert first.

What works well to ask

  • Describe the business domain and ask for a starter set of concepts and relationships.
  • Ask to add, rename, or remove specific concepts or domains.
  • Ask to connect concepts with a clear verb and relationship type.
  • Attach a domain brief (PDF/text) or relevant warehouse tables, then ask Copilot to propose a model grounded in that material.
Prefer business language (singular concept names). Copilot is guided not to invent attributes unless you ask for them.

Limits

  • Keep may require a plan that includes Agent Copilot.
  • Verb Suggest on the relationship form is a separate helper from Copilot threads.
  • Large conversations may need to be compacted when the thread is nearly full.
  • Resolve staged changes before starting another Agent edit.

How enrichment feeds data model generation

When you generate a data model, Snap Data Studio uses extra detail on your conceptual model to propose facts, dimensions, hubs, links, and related structures.

Event vs concept

Linked entities and roles

Link warehouse tables on concepts or relationships, then assign a role. Generation uses those links as follows: Links on concepts and on relationships both count. Relationship links are used when both connected concepts are included in generation.

Mapped vs unmapped attributes

Descriptions and synonyms

Synonyms, domains, and descriptions give Copilot and generation more context so the model is easier to understand. They are not strict matching rules during generation. For how generation fits into the wider workflow, continue in Create a logical & physical model.

What’s next