Coalesce

Overview

Coalesce is the visual, column-aware option in a category where everything else is code-first. You build pipelines as nodes on a graph, mapping columns in a grid and defining joins in an editor, rather than writing model files.

Underneath, it is metadata plus code generation. Each node type carries a specification, a create template that emits the DDL at deployment and a run template that emits the DML at refresh, both written in Jinja and SQL.

That template mechanism is what makes it more than a drag-and-drop tool. Teams codify their own patterns as user-defined node types, so a standard becomes a reusable component rather than a document nobody reads.

It is openly opinionated about modelling. Fact nodes implement Kimball fact tables and dimension nodes support both type 1 and type 2 slowly changing dimensions out of the box.

There is a code-first path too: SQL node types take inline annotations in plain SQL files, with columns and dependencies inferred from the query itself.

The product is now three things on one metadata layer, Transform for pipelines, Catalog for lineage, ownership and usage, and Quality for tests and monitoring, sharing one set of definitions and one governance model.

All compute runs in your warehouse. After deployment, refreshes execute entirely on your data platform and the data never leaves it.

The commercial model is worth understanding because it is unusual. Development is always free, you are charged only for actions in production, and every tier includes every product with no feature gates.

What differs by tier is scale, not capability: users, actions, projects, environments and enterprise controls such as single sign-on, private networking and a business associate agreement.

It is the only tool in this category with no open-source component at all, and no licence name is published because none applies.

Key features and capabilities

The same headings are used for every transformation & modelling entry, so two tools can be read side by side.

How transformations are written

• A visual, column-aware interface with a node graph, a mapping view and a join editor
• Node types come in two styles, configured visually or written as SQL with inline annotations
• SQL node types infer their columns and dependencies from the query itself
• Jinja and SQL templates generate the DDL at deploy and the DML at refresh
• User-defined node types turn an organisation's own patterns into reusable components

Execution and materialisation

• Built-in node types cover source, stage, persistent stage, fact, dimension and view
• Stage nodes truncate and reload by default; persistent stage retains data and tracks change history
• Dimension nodes support both type 1 current-state and type 2 historical tracking
• All compute runs on your own data platform, and data never leaves it
• Coalesce has no execution engine of its own; it generates and orchestrates SQL

Testing and data quality

• Quality is a first-class product, embedding tests and validations directly into pipelines
• Monitors are a billable unit alongside node executions, implying scheduled checks rather than run-time only
• Tests and versioned changes are part of the structured development model
• Unit tests, data contracts, freshness checks and anomaly detection are not published as named features

Environments and change management

• Projects and environments are explicit entities, limited by tier and unlimited on Enterprise
• All node types and nodes can be committed to Git and deployed to an environment
• Azure DevOps is documented as a supported Git provider
• A command-line tool and environment management API provide the continuous delivery hook
• Virtual environments, zero-copy clones and blue-green patterns are not published

Documentation and lineage

• Catalog generates lineage, ownership and usage metadata across data assets automatically
• Lineage comes from the shared metadata layer rather than being parsed after the fact
• An AI assistant in Catalog, with a documented context model
• A GraphQL API gives programmatic access to catalogue metadata
• Column-level lineage as a named feature, exposures and a metrics layer are not published

Platform coverage

• Snowflake, Databricks SQL, Microsoft Fabric and Google BigQuery
• Amazon Redshift is in private preview
• Node-type packages are published per platform through a marketplace
• Snowflake is the origin platform, with onboarding through Snowflake Partner Connect
• There is no local or embedded target

Where it runs and what it costs

• Managed software as a service only, with no self-hosted or local-only mode
• The command-line tool and APIs are automation clients against the hosted service
• Business Critical adds private networks, a business associate agreement and custom deployment
• Hosting regions are not published
• Billed per user plus per action, an action being a node execution, catalogue refresh or monitor refresh

Pricing

SubscriptionFree Developer tier; Starter $150 a user

Development is always free; you consume actions only when pipelines run in production. The Developer tier is $0 for one user and 2,000 actions a month, Starter $150 per user a month billed annually for up to four users and 15,000 actions, with Enterprise and Business Critical quote-only at 100,000 actions and unlimited projects. Every tier includes Transform, Catalog and Quality with no feature gates, which is unusual. Overage is a flat per-action rate with no tier jump, though that rate is not published. The trial runs 14 days.

Vendor pricing page →

Demos and videos

• Coalesce platform walkthrough in two minutes Product demo · YouTube
• Creating stage nodes in Coalesce Tutorial · YouTube

About Coalesce

Coalesce was founded in 2020 by Armon Petrossian and Satish Jayanthi, exited stealth in 2022 and is based in San Francisco. It is private and venture-backed, raising a $26m Series A led by Emergence Capital in 2022 and a $50m Series B in April 2024 led by Industry Ventures and Emergence, with Snowflake Ventures participating, for $81m in total. It publishes no customer count but names Caesars Entertainment, Denny's, TotalEnergies and the Tampa Bay Rays, and reported more than fourfold annual recurring revenue growth for the year to January 2024. The product is proprietary throughout.

Founded 2020 · San Francisco, California · coalesce.io