- Five routes away from dbt, side by side
- Keep the models, or keep the method?
- Running it yourself against paying for a cloud
- A one-hour bake-off with one real model
- Where a switch from dbt goes sideways
- How the bill is built
- Transformation layers and managed dbt
- Orchestrators and full pipeline platforms
- Where I'd point each kind of team
dbt alternatives fall into five types: SQL transformations inside one warehouse, managed transformation platforms, managed dbt environments, orchestrators that run the whole pipeline, and full pipeline platforms. Work out whether you want to keep your SQL models and change how they run, or replace the modelling layer itself, and you'll know which type to test.
Leaving dbt can mean three quite different projects. You might keep your SQL models and change where they're built and hosted, swap the modelling layer for a different one, or hand the whole pipeline to a platform that loads, transforms and schedules in one place. dbt Labs describes its product as a way to build reliable, governed data pipelines that speed up analytics and AI work, so whatever replaces it has to carry the governed part as well, not just run SQL.
Five routes away from dbt, side by side
| Type of tool | What changes | Fits | Check first |
|---|---|---|---|
| SQL transformations inside one warehouse | Your models move into the warehouse vendor's own tooling | A team whose data already lives in that warehouse | How tied you're happy to be to one warehouse |
| Managed transformation platform | The modelling layer, with lineage, testing and AI help built around it | A team that wants governance baked into how pipelines get made | How users, AI features or credits are billed |
| Managed dbt environment | Where dbt runs and how it's hosted, not how you write it | A team happy with its models but tired of running the tooling | Which data platforms its dbt adapter supports |
| Orchestrator | What schedules, runs and monitors every step | A pipeline with more moving parts than SQL alone | Whether you still need a modelling layer underneath |
| Full pipeline platform | Loading, transforming and running data in one product | A team that would rather buy one platform than join several | How much of your existing SQL can come across |
Read the table from the right-hand column. The check is the part that bites after a migration, and it's cheaper to answer it before you sign anything than after you've moved forty models.
Keep the models, or keep the method?
This is the fork that matters most, and it's worth being honest with yourself about where the pain sits. If your models, tests and documentation are in decent shape and what hurts is hosting, the editor or scheduling, you don't need to leave the dbt way of working at all. A managed dbt environment keeps your project and changes the plumbing. Datacoves, for example, wraps dbt in a hosted editor with Airflow alongside for orchestration.
If the pain is the modelling layer itself, the managed transformation platforms are where to look. The ones below pitch different things: governance and reusable building blocks, AI agents that build and run pipelines, or change tracking that only rebuilds the tables a change touches. Those are three different problems, so name yours before you book a demo.
And if every warehouse table you care about lives in one cloud, the warehouse vendor's own SQL tooling is the shortest hop. Dataform, for instance, runs SQL transformation pipelines inside BigQuery. The price of that convenience is that your transformation logic now sits inside one vendor's product.
For a wider look at the modelling side, there's a separate round-up of data transformation tools.
Running it yourself against paying for a cloud
The second fork is who runs the thing. Open-source tools you host move the bill from a subscription to servers and somebody's time: upgrades, security patches, access control and the night the scheduler falls over. Managed clouds take that on and charge for it. Neither is cheaper by default. It depends on whether you've got an engineer who'd rather run infrastructure or one who'd rather write models.
Orchestrators sit across this fork too: Airflow is a community-built Apache project, while Kestra calls itself an open-source orchestration platform. They schedule and monitor the whole pipeline, which is a bigger job than transformation, so check whether you'd still need a modelling layer underneath. On a managed plan of any type, ask about single sign-on, role-based access and audit logs before your security team does.
A one-hour bake-off with one real model
Demos are built to look smooth. Your own messy model is the better test. Pick one that has two parents and two children, plus at least one data test, and rebuild it in each candidate. Give yourself an hour per tool.
Start the clock and note how long it takes to get the first green run. Then change one column in a parent model and count what reruns: everything downstream, or only what the change touched? That one number tells you a lot about your future warehouse bill. Next, break the test on purpose. Where does the failure show up, how clear is the message, and who gets told? Open the lineage view and trace the final table back to its sources without asking anyone for help.
For an orchestrator, add a step: schedule the run, kill it halfway and see how the retry behaves. Keep your warehouse's query history open in another tab the whole time, because it shows you exactly what each tool sent to the warehouse, which no demo will.
Where a switch from dbt goes sideways
The first slip is swapping a modelling tool for a scheduler and expecting the same job done. An orchestrator will run your SQL, but it won't give you dbt's way of writing, testing and documenting models unless you bring a modelling layer along.
The second is starting without an inventory. Count your models, tests, macros and documentation, and list every dashboard and job that reads from them. That list decides how long the move takes far more than any feature does.
The third is buying on the AI demo. AI agents that build pipelines are a big part of the pitch on the managed platforms below, so test the plain SQL path as well, because that's the one your team will use on a bad day.
The last is cutting over in one go. Run old and new side by side on a handful of models, compare row counts and totals, and only switch once they match. If loading data is also on your list of problems, that's a separate decision, and there's a guide to data loading tools for it.
How the bill is built
Billing differs by type, so look at each vendor's pricing page rather than comparing headline figures. Coalesce prices per user, Paradime sells AI credits and Dagster shows a monthly plan; all three invite you to start free, and Matillion offers a free trial. Google calls Dataform a free service, with costs coming from the other Google Cloud services around it. For the rest, check the pricing page before you assume either way.
Whatever you pick, your warehouse bill keeps running underneath it, and that's where a transformation tool earns or loses its keep. The rerun count from the bake-off is your best early guide to it.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Transformation layers and managed dbt
Dataform
Google Cloud's service for developing and running scalable SQL transformation pipelines inside BigQuery, the nearest swap when BigQuery is already your warehouse.
Coalesce
A data transformation platform that turns business requirements into governed pipelines with any AI agent, using reusable node types so every pipeline comes out the same way.
Paradime
Paradime pitches production-ready AI agents that run your data platform, handling data requests and building pipelines, with connectors to land data from your sources.
Tobiko Cloud
Change-aware transformation: it tracks SQL changes and run history so only impacted tables rebuild, and makes development copies of production without extra warehouse processing.
Datacoves
Keeps dbt and changes the setup around it, as an enterprise dbt DataOps platform with hosted VS Code and Airflow for orchestrating dependencies.
Orchestrators and full pipeline platforms
Airflow
The community-built Apache platform for programmatically authoring, scheduling and monitoring workflows, with a modular design that uses a message queue to orchestrate workers.
Dagster
A data orchestrator for building, scheduling and monitoring pipelines, whose declarative automation lets you state the asset state you want while it handles the work.
Kestra
Open-source orchestration for infrastructure, IT operations, data and AI workflows, written in declarative YAML, run on schedules or events, with plugins for the rest.
Matillion
Data integration and ETL in one platform, loading your cloud warehouse on batch schedules through pre-built or custom connectors, plus Maia, which it calls an agentic data team.
Where I'd point each kind of team
A small analytics team whose models are fine but whose hosting isn't: stay with dbt and move to a managed dbt environment. It's the smallest change that fixes the real problem.
A team on one warehouse with no plans to add another: try the warehouse's own SQL tooling first. It's the shortest path, and you can always move later if the estate grows.
A team whose warehouse bill keeps climbing on rebuilds: test a change-aware transformation platform and judge it on the rerun count from your bake-off.
A data team with pipelines that reach well beyond SQL, into APIs, files and machine learning jobs: an orchestrator is the right centre of gravity, with your modelling layer kept underneath it. I'd pick open source if you've got someone who enjoys running it, and a managed plan if you haven't.
A business that wants one vendor to ring for loading, transforming and running data: a full pipeline platform, as long as your existing SQL can come across without a rewrite.
Do you make data transformation or orchestration tools? Tool makers can ask to be considered for a featured place on this list.
Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.


