Joins reference - Lightdash

Performance Best Practice: For optimal query performance, we recommend using wide tables wherever possible and minimising joins in the BI layer. While we offer advanced features like fanout protection to help with complex relationships, handling data transformations and complex logic directly in your SQL models will generally yield better performance than relying heavily on joins at query time. Consider pre-joining related data during your data modeling process rather than joining tables on-the-fly in dashboards and reports.

Adding joins in your models

Joins are defined at the same level as the model parameters in your YAML file.

models:
  - name: accounts
    config:
      meta:
        primary_key: id
        joins:
          - join: deals
            type: left
            sql_on: ${accounts.id} = ${deals.account_id}
            fields: [unique_deals, new_deals, won_deals, lost_deals, stage]
            relationship: one-to-many

models:
  - name: accounts
    meta:
      primary_key: id
      joins:
        - join: deals
          type: left
          sql_on: ${accounts.id} = ${deals.account_id}
          fields: [unique_deals, new_deals, won_deals, lost_deals, stage]
          relationship: one-to-many

type: model
name: accounts

primary_key: id

joins:
  - join: deals
    type: left
    sql_on: ${accounts.id} = ${deals.account_id}
    fields: [unique_deals, new_deals, won_deals, lost_deals, stage]
    relationship: one-to-many

When you open Lightdash, the dimensions and metrics from your joined model will appear below the ones in your selected model.

Omit fields to include all metrics and dimensions defined in the joined table.
When joined fields reference other fields, those other fields must be included.
The sql_on statement can include references to user attributes for row-level filtering.
Filters on a joined table’s fields, including dashboard filters, go in the query’s WHERE clause, after the join. To limit the rows read from the joined table, for example to a recent range of a partitioned table, put the condition in sql_on, with a parameter if users should pick the value. On a left join, base rows with no match inside that range still return, with nulls for the joined fields.

Rename a joined model

When joining tables, you may want to rename for readability. This can be done with the label tag, for example on the messages model it makes sense to rename the joined users table as “Sender”:

models:
  - name: messages
    config:
      meta:
        joins:
          - join: users
            label: Sender
            sql_on: ${messages.sent_by} = ${users.user_id}

models:
  - name: messages
    meta:
      joins:
        - join: users
          label: Sender
          sql_on: ${messages.sent_by} = ${users.user_id}

type: model
name: messages

joins:
  - join: users
    label: Sender
    sql_on: ${messages.sent_by} = ${users.user_id}

A join with no label and no alias shows the joined model’s own label. If several models join users this way, the table and its fields appear under the same name in every one of those explores, including in dashboard filters. Set a label on each join when the joined table plays a different role in each model. You can also override the description of the joined table by providing a description field:

models:
  - name: messages
    config:
      meta:
        joins:
          - join: users
            label: Sender
            description: User who sent the message
            sql_on: ${messages.sent_by} = ${users.user_id}

models:
  - name: messages
    meta:
      joins:
        - join: users
          label: Sender
          description: User who sent the message
          sql_on: ${messages.sent_by} = ${users.user_id}

type: model
name: messages

joins:
  - join: users
    label: Sender
    description: User who sent the message
    sql_on: ${messages.sent_by} = ${users.user_id}

Join the same table multiple times

If you need to join a table multiple times, you can use an alias to distinguish between the different tables. A common use case is joining a user table multiple times to another table depending on the type of user. For example this messages model has both a sender and a recipient:

models:
  - name: messages
    config:
      meta:
        joins:
          - join: users
            alias: sender
            sql_on: ${messages.sent_by} = ${sender.user_id}
          - join: users
            alias: recipient
            sql_on: ${messages.sent_to} = ${recipient.user_id}

models:
  - name: messages
    meta:
      joins:
        - join: users
          alias: sender
          sql_on: ${messages.sent_by} = ${sender.user_id}
        - join: users
          alias: recipient
          sql_on: ${messages.sent_to} = ${recipient.user_id}

type: model
name: messages

joins:
  - join: users
    alias: sender
    sql_on: ${messages.sent_by} = ${sender.user_id}
  - join: users
    alias: recipient
    sql_on: ${messages.sent_to} = ${recipient.user_id}

We recommend using alias values that are all lowercase and do not include spaces to avoid unexpected issues. Note the following important differences when using alias in joins:

You must reference the fields in the model using the new alias. Notice that the joins above use ${sender.user_id} rather than ${users.user_id}.
Because of the above, any fields in the base model or joined model that reference any field ${users.*} will fail to compile. Be careful of aliasing tables that are used in the base model.
Joined models are automatically relabelled with the alias but you may also customise this using the label: field as above.

Specify your join type

If you want to specify the type of join to be used in your SQL query, use the type field in your YAML configuration file. Set its value to one of the following: inner, left, right, or full. This will determine how the tables are joined in your query, aligning with SQL’s inner join, left outer join, right outer join, or full outer join respectively. Here’s an example of how to specify a join type:

models:
  - name: messages
    config:
      meta:
        joins:
          - join: users
            type: inner
            sql_on: ${messages.sent_by} = ${users.user_id}

models:
  - name: messages
    meta:
      joins:
        - join: users
          type: inner
          sql_on: ${messages.sent_by} = ${users.user_id}

type: model
name: messages

joins:
  - join: users
    type: inner
    sql_on: ${messages.sent_by} = ${users.user_id}

By default, if no type is specified, Lightdash will use a left join. Here’s a table to help you understand what each join type means and how it translates to SQL:

Join Type	Generated SQL	Description
inner	inner join	Returns rows that have matching values in both tables.
left	left outer join	Returns all rows from the left table, and the matching rows from the right table. Non-matching rows will have NULL for right table’s columns.
right	right outer join	Returns all rows from the right table, and the matching rows from the left table. Non-matching rows will have NULL for left table’s columns.
full	full outer join	Returns all rows when there is a match in either the left or right table records. Non-matching rows will have NULL for columns of the table that lacks a match.

inner

An inner join returns rows that have matching values in both tables. For example, if you have a users table and a subscriptions table, an inner join would return only the users who have a subscription. Here’s an example of how to specify an inner join:

models:
  - name: users
    config:
      meta:
        joins:
          - join: subscriptions
            sql_on: ${users.user_id} = ${subscriptions.user_id}
            type: inner

models:
  - name: users
    meta:
      joins:
        - join: subscriptions
          sql_on: ${users.user_id} = ${subscriptions.user_id}
          type: inner

type: model
name: users

joins:
  - join: subscriptions
    sql_on: ${users.user_id} = ${subscriptions.user_id}
    type: inner

left

A left join returns all rows from the left table (i.e. the model where you’re adding the join to), and the matching rows from the right table (i.e. the model you’ve specified in - join:). Non-matching rows will have NULL for right table’s columns. For example, if you have a users table and a subscriptions table, a left join would return all users, and the subscription information for users who have a subscription. Here’s an example of how to specify a left join:

models:
  - name: users
    config:
      meta:
        joins:
          - join: subscriptions
            sql_on: ${users.user_id} = ${subscriptions.user_id}
            type: left # you can omit this, as left is the default

models:
  - name: users
    meta:
      joins:
        - join: subscriptions
          sql_on: ${users.user_id} = ${subscriptions.user_id}
          type: left # you can omit this, as left is the default

type: model
name: users

joins:
  - join: subscriptions
    sql_on: ${users.user_id} = ${subscriptions.user_id}
    type: left # you can omit this, as left is the default

right

A right join returns all rows from the right table (i.e. the model you’ve specified in - join: ), and the matching rows from the left table (i.e. the model where you’re adding the join to). Non-matching rows will have NULL for left table’s columns. For example, if you have a users table and a subscriptions table, a right join would return all subscriptions, and the user information for users who have a subscription. Here’s an example of how to specify a right join:

models:
  - name: users
    config:
      meta:
        joins:
          - join: subscriptions
            sql_on: ${users.user_id} = ${subscriptions.user_id}
            type: right

models:
  - name: users
    meta:
      joins:
        - join: subscriptions
          sql_on: ${users.user_id} = ${subscriptions.user_id}
          type: right

type: model
name: users

joins:
  - join: subscriptions
    sql_on: ${users.user_id} = ${subscriptions.user_id}
    type: right

full

A full join returns all rows when there is a match in either the left or right table records. Non-matching rows will have NULL for columns of the table that lacks a match. For example, if you have a users table and a subscriptions table, a full join would return all users and all subscriptions, and the subscription information for users who have a subscription.

Defining join relationships

You can define the relationship between tables in your joins to help Lightdash show warnings and generate the appropriate SQL. This is especially useful for preventing SQL fanouts issues described in the SQL fanouts section. To define a relationship, add the relationship field to your join configuration.

models:
  - name: users
    config:
      meta:
        primary_key: user_id
        joins:
          - join: orders
            sql_on: ${users.user_id} = ${orders.user_id}
            relationship: one-to-many

models:
  - name: users
    meta:
      primary_key: user_id
      joins:
        - join: orders
          sql_on: ${users.user_id} = ${orders.user_id}
          relationship: one-to-many

type: model
name: users

primary_key: user_id

joins:
  - join: orders
    sql_on: ${users.user_id} = ${orders.user_id}
    relationship: one-to-many

Make sure that you consider the direction of the join when defining the relationship. If you incorreclty define the join relationship, your will be affected by fanouts.

The following join relationships are supported:

one-to-many - Starting table has 1 record, joined table has many matches
many-to-one - Starting table has many records, joined table has 1 match
one-to-one - Starting table has 1 record, joined table has 1 match
many-to-many - Multiple records in the starting table match multiple records in the joined table

Helpful Steps for Determining Join Relationships

Step 1: Identify your starting table

Which table are you joining FROM? Direction matters: Accounts joining to Users (one-to-many) is completely different from users joining to accounts (many-to-one), even though it’s the same data.

Step 2: Count the expected matches and name the join relationship

For any record in your starting table, ask: “How many matching records will I find in the table I’m joining to and vice versa?” Refer to the supported join relationships listed above.The examples below detail some more complex join relationships:

Chained Join Example

Don’t try to figure out Accounts → Users → Tracks all at once. Analyze each join separately:

First: Accounts → Users (one-to-many)
Then: Users → Tracks (one-to-many)
Overall result: Accounts → Tracks (one-to-many)

The accounts.yml file will look like this:

  models:
    - name: accounts
      meta:
        primary_key: account_id
        description: List of all customer and prospective customer Accounts pulled from our CRM
        joins:
          - join: users
            relationship: one-to-many
            sql_on: ${accounts.account_id} = ${users.account_id}
            type: left
          - join: tracks
            relationship: one-to-many
            sql_on: ${users.user_id} = ${tracks.user_id}
            type: left

  models:
    - name: accounts
      config:
        meta:
          primary_key: account_id
          description: List of all customer and prospective customer Accounts pulled from our CRM
          joins:
            - join: users
              relationship: one-to-many
              sql_on: ${accounts.account_id} = ${users.account_id}
              type: left
            - join: tracks
              relationship: one-to-many
              sql_on: ${users.user_id} = ${tracks.user_id}
              type: left

type: model
name: accounts

primary_key: account_id
description: List of all customer and prospective customer Accounts pulled from our CRM

joins:
  - join: users
    relationship: one-to-many
    sql_on: ${accounts.account_id} = ${users.account_id}
    type: left
  - join: tracks
    relationship: one-to-many
    sql_on: ${users.user_id} = ${tracks.user_id}
    type: left

The above setup will consider both Accounts and Users as being susceptible to fanouts and these would be handled accordingly. When you chain two one-to-many relationships, you get a one-to-many relationship from your starting table to your final table (Accounts can have many Tracks).Note that if I wanted to join Users and Accounts onto the Tracks, where Tracks is the starting model, the direction of the relationship would look different:The tracks.yml model would look like this:

  models:
    - name: tracks
      meta:
        primary_key: account_id
        description: List of all customer and prospective customer Accounts pulled from our CRM
        joins:
          - join: users
            relationship: many-to-one
            sql_on: ${users.user_id} = ${tracks.user_id}
            type: right
          - join: accounts
            relationship: many-to-one
            sql_on: ${users.account_id} = ${accounts.account_id}
            type: right

  models:
    - name: tracks
      config:
        meta:
          primary_key: account_id
          description: List of all customer and prospective customer Accounts pulled from our CRM
          joins:
            - join: users
              relationship: many-to-one
              sql_on: ${users.user_id} = ${tracks.user_id}
              type: right
            - join: accounts
              relationship: many-to-one
              sql_on: ${users.account_id} = ${accounts.account_id}
              type: right

type: model
name: tracks

primary_key: account_id
description: List of all customer and prospective customer Accounts pulled from our CRM

joins:
  - join: users
    relationship: many-to-one
    sql_on: ${users.user_id} = ${tracks.user_id}
    type: right
  - join: accounts
    relationship: many-to-one
    sql_on: ${users.account_id} = ${accounts.account_id}
    type: right

Complex Join Example

We want to see all Accounts and all Deals, but we only want to see Users (and their associated event Tracks) for Accounts that have at least one Deal in the “Won” stage.This join example is covered in our Demo site here.This requires a complex join that involves 4 different tables.

First: Accounts → Deals (one-to-many)
Next: Accounts and Deals → Users (many-to-many) - each Account+ Deal combination can be associated with many Users and each user can be associated with multiple Deals.
Then: Users → Tracks (one-to-many)

A normal SQL join that does not account for fanouts would look like this:

select 
  * 
from  
  accounts 
  left join deals on 
    accounts.account_id = deals.account_id
  left join users on 
    accounts.account_id = users.account_id and deals.stage ='Won'
  left join tracks on 
    users.user_id = tracks.user_id

And the accounts.yml would look like this:

  models:
    - name: accounts
      meta:
        primary_key: account_id
        description: List of all customer and prospective customer Accounts pulled from our CRM
        joins:
          - join: deals
            relationship: one-to-many
            sql_on: ${accounts.account_id} = ${deals.account_id}
            type: left
          - join: users
            relationship: many-to-many
            sql_on: ${accounts.account_id} = ${users.account_id} and ${deals.stage} = 'Won'
            type: left
          - join: tracks
            relationship: one-to-many
            sql_on: ${users.user_id} = ${tracks.user_id}
            type: left

  models:
    - name: accounts
      config:
        meta:
          primary_key: account_id
          description: List of all customer and prospective customer Accounts pulled from our CRM
          joins:
            - join: deals
              relationship: one-to-many
              sql_on: ${accounts.account_id} = ${deals.account_id}
              type: left
            - join: users
              relationship: many-to-many
              sql_on: ${accounts.account_id} = ${users.account_id} and ${deals.stage} = 'Won'
              type: left
            - join: tracks
              relationship: one-to-many
              sql_on: ${users.user_id} = ${tracks.user_id}
              type: left

type: model
name: accounts

primary_key: account_id
description: List of all customer and prospective customer Accounts pulled from our CRM

joins:
  - join: deals
    relationship: one-to-many
    sql_on: ${accounts.account_id} = ${deals.account_id}
    type: left
  - join: users
    relationship: many-to-many
    sql_on: ${accounts.account_id} = ${users.account_id} and ${deals.stage} = 'Won'
    type: left
  - join: tracks
    relationship: one-to-many
    sql_on: ${users.user_id} = ${tracks.user_id}
    type: left

In this case, the fanout protection logic will consider metrics from all models to be susceptible to fanouts.

Step 3: Check for conditional joins

Look for any AND conditions in your join logic (like and ${deals.stage} = 'Won'). These can change your relationship from what you’d expect - a typical one-to-many might become many-to-many when you add conditions.

Step 4: Validate with sample data

Pick one record from your starting table and manually trace through the joins. Count how many final records you get - this helps catch relationship mistakes before they cause problems.

Always join a table

If you need a table to always be joined, you can set the always field to true.

models:
  - name: messages
    config:
      meta:
        joins:
          - join: users
            sql_on: ${messages.sent_by} = ${users.user_id}
            always: true

models:
  - name: messages
    meta:
      joins:
        - join: users
          sql_on: ${messages.sent_by} = ${users.user_id}
          always: true

type: model
name: messages

joins:
  - join: users
    sql_on: ${messages.sent_by} = ${users.user_id}
    always: true

This will make sure that even when you don’t select any of the fields from the joined table it will still be joined in the query.

Only select a subset of fields from a join

Use the fields tag to select a subset of fields from a join. This is useful if you want to join a model but only a few of its fields are useful in the joined context. For example this messages model only needs the name andemail fields from the users model. Note we must also include the user_id field since it’s needed for the join.

models:
  - name: messages
    config:
      meta:
        joins:
          - join: users
            sql_on: ${messages.sent_by} = ${users.user_id}
            fields: [user_id, email, name]

models:
  - name: messages
    meta:
      joins:
        - join: users
          sql_on: ${messages.sent_by} = ${users.user_id}
          fields: [user_id, email, name]

type: model
name: messages

joins:
  - join: users
    sql_on: ${messages.sent_by} = ${users.user_id}
    fields: [user_id, email, name]

To take all but a few fields, list the fields once in a set on the joined model and exclude the ones you don’t want with -, for example fields: [user_fields*, -signup_source, -region]. There’s no built-in set of every field, so the set lists its fields itself.

Using joined dimensions or metrics in your YAML

Once you’ve joined a model, you can reference the metrics and dimensions from your joined model in your configurations. For example, I can filter one of my metrics using a dimension from my joined model, like this:

models:
  - name: users
    config:
      meta:
        joins:
          - join: web_sessions
            sql_on: ${web_sessions.user_id} = ${users.user_id}
          - join: subscriptions
            sql_on: ${subscriptions.user_id} = ${users.user_id} AND ${subscriptions.is_active}

    columns:
      - name: user_id
        config:
          meta:
            metrics:
              num_unique_premium_user_ids:
                type: count_distinct
                filters:
                  - subscriptions.plan: premium

models:
  - name: users
    meta:
      joins:
        - join: web_sessions
          sql_on: ${web_sessions.user_id} = ${users.user_id}
        - join: subscriptions
          sql_on: ${subscriptions.user_id} = ${users.user_id} AND ${subscriptions.is_active}

    columns:
      - name: user_id
        meta:
          metrics:
            num_unique_premium_user_ids:
              type: count_distinct
              filters:
                - subscriptions.plan: premium

type: model
name: users

joins:
  - join: web_sessions
    sql_on: ${web_sessions.user_id} = ${users.user_id}
  - join: subscriptions
    sql_on: ${subscriptions.user_id} = ${users.user_id} AND ${subscriptions.is_active}

dimensions:
  - name: user_id

metrics:
  num_unique_premium_user_ids:
    type: count_distinct
    sql: ${TABLE}.user_id
    filters:
      - subscriptions.plan: premium

You can also reference joined metrics and dimensions in custom sql, like this:

models:
  - name: users
    config:
      meta:
        joins:
          - join: web_sessions
            sql_on: ${web_sessions.user_id} = ${users.user_id}
          - join: subscriptions
            sql_on: ${subscriptions.user_id} = ${users.user_id} AND ${subscriptions.is_active}

    columns:
      - name: user_id
        config:
          meta:
            dimension:
              sql: IF(${subscriptions.plan} IS NULL, NULL, ${user_id})
            metrics:
              num_unique_premium_user_ids:
                type: count_distinct
                sql: IF(${subscriptions.plan} = 'premium', ${user_id}, NULL)

models:
  - name: users
    meta:
      joins:
        - join: web_sessions
          sql_on: ${web_sessions.user_id} = ${users.user_id}
        - join: subscriptions
          sql_on: ${subscriptions.user_id} = ${users.user_id} AND ${subscriptions.is_active}

    columns:
      - name: user_id
        meta:
          dimension:
            sql: IF(${subscriptions.plan} IS NULL, NULL, ${user_id})
          metrics:
            num_unique_premium_user_ids:
              type: count_distinct
              sql: IF(${subscriptions.plan} = 'premium', ${user_id}, NULL)

type: model
name: users

joins:
  - join: web_sessions
    sql_on: ${web_sessions.user_id} = ${users.user_id}
  - join: subscriptions
    sql_on: ${subscriptions.user_id} = ${users.user_id} AND ${subscriptions.is_active}

dimensions:
  - name: user_id
    sql: IF(${subscriptions.plan} IS NULL, NULL, ${user_id})

metrics:
  num_unique_premium_user_ids:
    type: count_distinct
    sql: IF(${subscriptions.plan} = 'premium', ${user_id}, NULL)

If you get a has a metric reference ... which matches no metric error, see Using custom SQL in non-aggregate metrics. Every joined field that you reference in a YAML file adds a dependency that you’ll have to include in all future references to that model. For example, you might define deals.unique_enterprise_deals by using a joined field in a filter: ${accounts.segment} = 'Enterprise'. Then later you need to join deals to a marketing_attribution model. The unique_enterprise_deals metric must be excluded from the join unless you also join accounts to get the accounts.segment field. Check out our dimensions and metrics reference docs to see all of the other configurations you can use with your joined fields.

Joins are not transitive

Joins defined on one model are not automatically available when that model is joined to another model. Each explore only knows about the joins explicitly defined on the model you are exploring from. For example, if accounts joins deals and users joins accounts, the deals join is not carried over to the users explore. Any dimension or metric on accounts that references a deals field (like ${deals.stage}) will fail to compile when exploring from users.

users → accounts          ✅ works
users → accounts → deals  ❌ not automatic

This commonly surfaces when you define a metric on one model using a dimension from a joined table, and then join that model elsewhere. The metric references a field that doesn’t exist in the new context.

How to resolve cross-join field references

There are three approaches, depending on your use case: Option 1: Materialize the field in dbt (recommended) Compute the value as a real column in your dbt model so it has no cross-model dependency. This is the cleanest approach and avoids the issue entirely.

-- in your dbt model for accounts
SELECT
  *,
  (SELECT COUNT(*) FROM deals WHERE deals.account_id = accounts.id AND deals.stage = 'Won') AS won_deals_count
FROM accounts

Option 2: Add a chained join to the other explore Explicitly add the missing table as a chained join on any explore that needs it. This makes the referenced fields available in that context.

models:
  - name: users
    config:
      meta:
        joins:
          - join: accounts
            sql_on: ${users.account_id} = ${accounts.id}
          - join: deals
            sql_on: ${accounts.id} = ${deals.account_id}

models:
  - name: users
    meta:
      joins:
        - join: accounts
          sql_on: ${users.account_id} = ${accounts.id}
        - join: deals
          sql_on: ${accounts.id} = ${deals.account_id}

type: model
name: users

joins:
  - join: accounts
    sql_on: ${users.account_id} = ${accounts.id}
  - join: deals
    sql_on: ${accounts.id} = ${deals.account_id}

Option 3: Exclude the dependent metrics with fields Use the fields tag on the join to omit metrics that depend on unavailable tables. The metric won’t be accessible in the explore, but the error is suppressed. Setting hidden: true on the metric doesn’t do this: hidden fields are still compiled, so the error remains.

models:
  - name: users
    config:
      meta:
        joins:
          - join: accounts
            sql_on: ${users.account_id} = ${accounts.id}
            fields: [id, name]  # omit metrics that reference deals

models:
  - name: users
    meta:
      joins:
        - join: accounts
          sql_on: ${users.account_id} = ${accounts.id}
          fields: [id, name]  # omit metrics that reference deals

type: model
name: users

joins:
  - join: accounts
    sql_on: ${users.account_id} = ${accounts.id}
    fields: [id, name]  # omit metrics that reference deals

Hide joined tables

Sometimes, you need to use an intermediate model to join two models together and don’t want to show its columns in the list of available metrics/dimensions. You can add a hidden: true tag to joins, like this to hide these intermediate models:

models:
  - name: users
    config:
      meta:
        joins:
          - join: map_users_organizations
            sql_on: users.user_id = map_users_organizations.user_id
            hidden: true
          - join: organizations
            sql_on: organizations.organization_id = map_users_organizations.organization_id

models:
  - name: users
    meta:
      joins:
        - join: map_users_organizations
          sql_on: users.user_id = map_users_organizations.user_id
          hidden: true
        - join: organizations
          sql_on: organizations.organization_id = map_users_organizations.organization_id

type: model
name: users

joins:
  - join: map_users_organizations
    sql_on: users.user_id = map_users_organizations.user_id
    hidden: true
  - join: organizations
    sql_on: organizations.organization_id = map_users_organizations.organization_id

Using just the hidden tag in the join and changing nothing else will remove the joined table from the sidebar of metrics/dimensions in users, but it will still appear in the Tables list as a table to explore on its own.

Two sidebars side by side: the hidden joined table is absent from dimensions and metrics but still listed under Tables

You can adjust which tables show up in the Tables list with Tables configuration in your project settings.

SQL fanouts

When joining tables in SQL, it can accidentally inflate your metrics if those tables have one-to-many relationships. Understanding when this happens helps you know which metrics need special handling. When dealing with joins that might cause metric inflation, certain types of calculations remain safe to use.

Safe vs. risky metrics

These metrics stay accurate even when the data contains duplicates:

min and max: min(order_total) and max(order_total) give you the same answer whether you have duplicates or not, because they’re looking for the extreme values, not counting or adding things up. The maximum value in a set doesn’t change just because that value appears multiple times.
Aggregate functions with distinct: count(distinct user_id) ensures each user is counted only once, regardless of how many times they appear in the joined result.

These metrics can give you the wrong answers:

sum, count, avg without using distinct
Custom calculations that treat duplicated rows as separate data points

Common scenarios and what gets affected

Single one-to-many join

Example: When you join an organization table to a users table, each organization can have multiple users.In this scenario:✅ User metrics remain accurate Since each user appears exactly once in the joined result, any metrics calculated on the user level (like average user age or total user count) will be correct.❌ Organization metrics become inflated - Each organization will appear multiple times in the result set (once for each associated user). This means organization-level metrics will be counted multiple times, leading to inflated results.For example, if “ABC Corp” has 2 users, its org_total_users metric cannot be summed because it will be duplicated twice.Raw tablesorganizations table

organization_id	organization_name	org_total_users
1	ABC Corp	2
2	XYZ Ltd	3

users table

organization_id	user_id	user_age
1	100	57
1	200	13
2	300	20
2	400	30
2	500	19

Joined OutputOrganizations are multiplied by the number of users

user_id	organization_id	org_total_users	user_age
100	1	2	57
200	1	2	13
300	2	3	20
400	2	3	30
500	2	3	19

We cannot sum org_total_users on this joined table because it will be incorrect.

Chained one-to-many joins

Example: When you create a chain of one-to-many relationships by joining users to their orders, and then orders to their support tickets.In this scenario:✅ Ticket metrics remain accurate - The final table (tickets) in the chain will have accurate metrics because each ticket appears exactly once in the result set.❌ Upstream tables experience compounding inflation - The further up the chain you go, the worse the inflation gets:

Order metrics become inflated by the number of tickets per order. If Order #1001 has 3 support tickets, any metrics for this order (like order value) will be counted 3 times.
User metrics suffer even more inflation as they’re multiplied by the total number of tickets. If a user has 2 orders with 3 tickets each, user-level metrics (like lifetime value) will be counted 6 times in aggregate calculations.

For example, if the user Alice has 2 orders and each of these orders have 3 tickets, after joining users → orders → tickets, the result would look like:Raw tablesusers table

user_id	user_name	user_age
1	Alice	23

orders table

order_id	user_id	order_total_items
1001	1	5
2001	1	2

tickets Table

order_id	ticket_id	ticket_time_to_first_response_mins
1001	A	8
1001	B	62
1001	C	47
2001	D	5
2001	E	1
2001	F	89

Joined outputStep 1: First Join users to orders

user_name	user_age	order_id	order_total_items
Alice	23	1001	5
Alice	23	2001	2

Step 2: Join the result of step 1 to tickets

user_name	order_id	ticket_id	user_age	order_total_items	ticket_time_to_first_response_mins
Alice	1001	A	23	5	8
Alice	1001	B	23	5	62
Alice	1001	C	23	5	47
Alice	2001	D	23	2	5
Alice	2001	E	23	2	1
Alice	2001	F	23	2	89

Note that users and orders are duplicated but tickets are not, which means we can only safely apply aggregates to metrics on the tickets table.

Multiple one-to-many joins

Example: When you join users → orders AND users → tickets (both orders and tickets relate to users, but not directly to each other)This creates the most problematic data situation, when you join multiple one-to-many relationships that branch from the same table but aren’t directly related to each other. In this scenario:❌ EVERY aggregate metric gets inflated due to the cartesian product effect

User metrics become massively inflated by the product of orders × tickets. If a user has 5 orders and 10 support tickets, each user-level metric will be counted 50 times!
Order metrics get inflated by the number of tickets the user has.
Ticket metrics get inflated by the number of orders the user has.

For example, if the user Alice has 2 orders and 3 tickets, after joining users to both orders and tickets separately, the result would look like:Raw tablesusers table

user_id	user_name	user_credit_amount
1	Alice	100

orders table

user_id	order_id	order_total_items
1	1001	5
1	2001	2

tickets Table

user_id	ticket_id	ticket_time_to_first_response_mins
1	A	8
1	B	62
1	C	47

The cartesian product - (every record is joined to every other record) of users, orders and tickets.

user_name	order_id	ticket_id	user_credit_amount	order_total_items	ticket_time_to_first_response_mins
Alice	1001	A	100	5	8
Alice	1001	B	100	5	62
Alice	1001	C	100	5	47
Alice	2001	A	100	2	8
Alice	2001	B	100	2	62
Alice	2001	C	100	2	47

Note that there is duplication in users, orders and tickets records. We cannot apply aggregates on any metrics without deduplicating records. This creates the worst-case scenario where no aggregate metric (sum, count, avg) is safe without explicit deduplication techniques.

Handling fanouts

Lightdash can automatically handle deduplicating metrics that are inflated due to fanouts if join relationships are specified in YAML as described in Defining join relationships, above. For example:

models:
  - name: users
    config:
      meta:
        primary_key: user_id
        joins:
          - join: organizations
            sql_on: ${organizations.organization_id} = ${users.organization_id}
            relationship: many-to-one
          - join: orders
            sql_on: ${users.user_id} = ${orders.user_id}
            relationship: one-to-many

models:
  - name: users
    meta:
      primary_key: user_id
      joins:
        - join: organizations
          sql_on: ${organizations.organization_id} = ${users.organization_id}
          relationship: many-to-one
        - join: orders
          sql_on: ${users.user_id} = ${orders.user_id}
          relationship: one-to-many

type: model
name: users

primary_key: user_id

joins:
  - join: organizations
    sql_on: ${organizations.organization_id} = ${users.organization_id}
    relationship: many-to-one
  - join: orders
    sql_on: ${users.user_id} = ${orders.user_id}
    relationship: one-to-many

Specifying a primary_key and the join relationship allows Lightdash to:

Understand the intended row-level granularity
Prevent metric inflation by deduplicating joined rows

Once you’ve included primary keys and a join relationship, Lightdash will add CTEs to the compiled SQL query that ensure metrics are not inflated.

How Lightdash solves SQL fanouts

Lightdash uses a pattern of Common Table Expressions (CTEs) to solve the fanout problem. Here’s how it works:

cte_keys: Contains dimensions (like payment_method, order_id) that define the grain of your final results and the primary keys. Any field you want to GROUP BY in your final output should be included here.
cte_metrics: Performs calculations on metrics while maintaining the correct grain established by the keys CTE. This prevents double-counting when aggregating across related tables.
cte_unaffected: Calculates all metrics that are not affected by fanouts. This includes metrics that exclude duplicates by definition (i.e. MIN(), MAX() and COUNT(DISTINCT)) as well as metrics that are calculated on the table that is on the many side of an one-to-many or many-to-one join relationship. For example, if you have joined accounts to deals using a one-to-many join relationship, SUM(deals.amount) would be calculated in this unaffected_cte because the deals data is not susceptible to fanouts.
final: Join the metrics CTEs together to create the complete result set.

Lightdash creates a separate cte_keys and cte_metrics for each table that contains metrics with fanouts. This is why you’ll see names like “cte_keys_orders” and “cte_metrics_orders” in the example below, indicating they’re specific to the orders table.

Examples

# Step 1: Create cte_keys that determine the final grain of your results i.e. whatever we will group by.
# Exclude fields that we use in aggregations to calculate metrics e.g. `"orders".amount`
WITH cte_keys_orders AS (
    SELECT DISTINCT
      "orders".status AS "orders_status", -- grouping dimension
      "orders".order_id AS "pk_order_id"  -- primary key
    FROM "postgres"."jaffle"."payments" AS "payments"
    LEFT OUTER JOIN "postgres"."jaffle"."orders" AS "orders"
      ON ("orders".order_id) = ("payments".order_id)
),
# Step 2: Calculate metrics that are affected by fanouts
cte_metrics_orders AS (
    SELECT
      cte_keys_orders."orders_status",
      SUM("orders".amount) AS "orders_total_order_amount" -- order metric (affected by fanout)
    FROM cte_keys_orders
    LEFT JOIN "postgres"."jaffle"."orders" AS "orders" ON cte_keys_orders."pk_order_id" = "orders".order_id -- join with primary keys
    GROUP BY 1 -- Note orders_status are grouping dimensions
),
# Step 3: Calculate metrics that are not affected by fanouts
cte_unaffected AS (
    SELECT
      "orders".status AS "orders_status",
      COUNT(DISTINCT "payments".payment_id) AS "payments_unique_payment_count" -- payment metric (NOT affected by fanout)
    FROM "postgres"."jaffle"."payments" AS "payments"
    LEFT OUTER JOIN "postgres"."jaffle"."orders" AS "orders"
      ON ("orders".order_id) = ("payments".order_id)
    GROUP BY 1 -- Note orders_status are grouping dimensions
)
# Step 4: Join the metrics CTEs together to create the final result with properly calculated metrics
SELECT
  cte_unaffected.*,
  cte_metrics_orders."orders_total_order_amount" AS "orders_total_order_amount"
FROM cte_unaffected
INNER JOIN cte_metrics_orders ON (
  cte_unaffected."orders_status" = cte_metrics_orders."orders_status" OR ( cte_unaffected."orders_status" IS NULL AND cte_metrics_orders."orders_status" IS NULL )
)
ORDER BY "orders_total_order_amount" DESC
LIMIT 500

Known limitations

There are a few situations where Lightdash doesn’t currently handle inflated metrics:

Two repeated columns unnested together

For the Repeated columns … are unnested together query warning, see Nested and repeated columns.

Metrics that reference a dimension from joined table

When a metric in one table references a dimension from a joined table, Lightdash’s fanout handling may not correctly deduplicate the results.ExampleReferences in metric sql

metrics:
  total_amount:
    type: sum
    sql: ${amount}+${expenses.amount}

RecommendationTo avoid this limitation, consider moving this logic to your dbt models. By transforming your data in dbt first, you can pre-calculate these dimensions without needing to reference dimensions from joined tables in Lightdash.

Metrics that have a filter with a dimension from joined table

Metrics that include filters based on dimensions from joined tables can also cause issues with Lightdash’s fanout handling.Example

metrics:
  count_users:
    type: count_distinct
    filters:
      - web_sessions.is_bot_user: false

RecommendationTo avoid this limitation, consider moving this logic to your dbt models. By applying these filters in your dbt transformations, you can create pre-filtered metrics that don’t require referencing dimensions from joined tables in Lightdash.

Metrics that have multiple levels of references

Metrics that involve multiple levels of references or dependencies can be particularly challenging for Lightdash’s fanout handling.Example

# orders.yml
metrics:
  total_amount:
    type: number
    sql: ${amount}+${expenses.total_amount}

# expenses.yml
metrics:
  total_amount:
    type: number
    sql: ${amount}+${tariffs.amount}

RecommendationTo avoid this limitation, consider moving this complex logic to your dbt models. By pre-calculating these metrics in dbt, you can flatten the dependency chain and avoid the need for multi-level references between joined tables in Lightdash.

Rolledup metrics

Rolledup metrics are pre-aggregated metrics that have already been calculated at a specific granularity in your data warehouse. When these metrics are used in queries that involve joins creating fanouts, they can become inflated because the pre-aggregated values get duplicated across the fanned-out rows.Example: Orders and payments analysisConsider a scenario where you have an orders table with total_amount and a payments table with payment methods. For each order, you can have multiple payments with different methods:orders table:

order_id	total_amount
1001	100.00
1002	250.00

payments table:

payment_id	order_id	payment_method	amount
1	1001	cash	60.00
2	1001	card	40.00
3	1002	card	250.00

If you select payment method and average orders total_amount, the results are wrong because total_amount is a rolledup metric that can’t be split by method:

payment_method	avg_order_total
cash	100.00
card	175.00

This is incorrect because:

The cash average should reflect that cash was only used for part of order 1001 ($100), but the query shows $100 as if cash paid for the entire order
The card average shows $175 (average of $100 and $250), but this doesn’t represent the actual relationship between card payments and order totals

The issue is that total_amount is a rolledup metric at the order level, but when joined with payments, it gets duplicated across payment methods, making it impossible to correctly analyze the relationship between payment methods and order totals.Best practices for rolledup metricsTo avoid issues with rolledup metrics in joins that create fanouts:

Avoid rolledup metrics when possible: Instead of using pre-aggregated values, use the underlying detail-level data. For example, instead of using a pre-calculated total_amount at the order level, use individual payment amounts that can be properly aggregated.
Name rolledup metrics clearly: If you must use rolledup metrics, give them descriptive names that indicate their pre-aggregated nature and limitations.
Provide clear descriptions: Always include detailed descriptions in your dbt model’s YAML that explain the metric’s granularity and any limitations when used with joins.

    models:
      - name: orders
        columns:
          - name: total_amount
            meta:
              metrics:
                total_order_amount:
                  type: sum
                  description: 'Pre-aggregated total amount per order. Cannot be meaningfully split by payment method or other transaction-level dimensions.'

    models:
      - name: orders
        columns:
          - name: total_amount
            config:
              meta:
                metrics:
                  total_order_amount:
                    type: sum
                    description: 'Pre-aggregated total amount per order. Cannot be meaningfully split by payment method or other transaction-level dimensions.'

type: model
name: orders

dimensions:
  - name: total_amount

metrics:
  total_order_amount:
    type: sum
    sql: ${TABLE}.total_amount
    description: 'Pre-aggregated total amount per order. Cannot be meaningfully split by payment method or other transaction-level dimensions.'

Intentional fanouts

In some business scenarios, fanouts are actually desired and represent real data relationships. A common example is when you need to calculate per-user fees or charges.Example: Per-user billingConsider a scenario with accounts, users, and per-user fees where different accounts pay different rates:accounts table:

account_id	account_name
1	TechCorp
2	StartupXYZ

users table:

user_id	account_id	user_name
101	1	Alice
102	1	Bob
201	2	David

fees table:

account_id	fee_type	per_user_amount
1	Support Fee	80.00
1	Training Fee	120.00
2	Support Fee	50.00
2	Training Fee	75.00

When you join these tables:

SELECT 
  a.account_name, 
  u.user_name, 
  f.fee_type, 
  f.per_user_amount
FROM 
  accounts a
  LEFT JOIN users u ON a.account_id = u.account_id
  LEFT JOIN fees f ON a.account_id = f.account_id;

The resulting fanout is exactly what you want for billing:

account_name	user_name	fee_type	per_user_amount
TechCorp	Alice	Support Fee	80.00
TechCorp	Alice	Training Fee	120.00
TechCorp	Bob	Support Fee	80.00
TechCorp	Bob	Training Fee	120.00
StartupXYZ	David	Support Fee	50.00
StartupXYZ	David	Training Fee	75.00

In this case, each row represents a real charge, and the total billing (TechCorp pays $400, StartupXYZ pays $125) is correctly calculated by summing all rows.Handling intentional fanoutsLightdash does not currently provide a built-in way to explicitly handle intentional fanouts. Using intentional fanouts can cause issues when defining other table relationships and may conflict with Lightdash’s automatic fanout handling.We recommend always use fanout protection for joins. When you actually want a fanout (like per-user billing), create a dedicated dbt model that handles the logic and gets your data to the right granularity first. Then use that model in Lightdash.

​Adding joins in your models

​Rename a joined model

​Join the same table multiple times

​Specify your join type

​inner

​left

​right

​full

​Defining join relationships

The following join relationships are supported:

​Step 1: Identify your starting table

​Step 2: Count the expected matches and name the join relationship

Chained Join Example

Complex Join Example

​Step 3: Check for conditional joins

​Step 4: Validate with sample data

​Always join a table

​Only select a subset of fields from a join

​Using joined dimensions or metrics in your YAML

​Joins are not transitive

​How to resolve cross-join field references

​Hide joined tables

​SQL fanouts

​Safe vs. risky metrics

​Common scenarios and what gets affected

​Handling fanouts

​How Lightdash solves SQL fanouts

Examples

​Known limitations

Adding joins in your models

Rename a joined model

Join the same table multiple times

Specify your join type

inner

left

right

full

Defining join relationships

Step 1: Identify your starting table

Step 2: Count the expected matches and name the join relationship

Step 3: Check for conditional joins

Step 4: Validate with sample data

Always join a table

Only select a subset of fields from a join

Using joined dimensions or metrics in your YAML

Joins are not transitive

How to resolve cross-join field references

Hide joined tables

SQL fanouts

Safe vs. risky metrics

Common scenarios and what gets affected

Handling fanouts

How Lightdash solves SQL fanouts

Known limitations