How to Prevent Unintentional Joins in Looker Explore?

[shakthydoss]

Looker's Explore automatically includes any joins defined in the Explore block. While this simplifies query building, it also leads to unintended joins when we only need data from a single table.

For example, if I want to get a simple distinct count or list from a particular column in table 'B', the Explore still joins with table 'A', leading to inaccurate results. I have several columns in table 'B' where I might need a distinct count or list, but these unnecessary joins affect the data.

I understand that this happens because of the join configurations defined in our Explore file, which are absolutely necessary for other queries.

explore: table_a {
  join: table_b {
  type: left_outer
  sql_on: ${table_a.a_id} = ${table_b.b_id} ;;
  relationship: many_to_one
  }

  join: table_c {
  type: left_outer
  sql_on: ${table_a.a_id} = ${table_c.c_id} ;;
  relationship: many_to_one
  }
}

My Questions:
Is there a better way or a workaround to prevent these unintended joins?
If yes, how can I apply the same workaround to other tables like table_c and table_d, where joins are required for some queries but not always?

Impact on Explore Assistance:
Since table_a Explore is linked to Explore Assistance, would introducing multiple Explores for individual tables (e.g., separate Explores for table_b, table_c, etc.) affect the behaviour of Explore Assistance?

Would appreciate any guidance on handling this more efficiently. Thanks!

[sam8]

Whenever you start a query from an Explore, the Explore's base table will always be included in the query.

So, if you want to start a query from table B without any joins, you would have to create a new Explore starting from table B.

I'm not 100% following your question about Explore Assistance, but there should be no problem with referencing the same table in multiple Explores. Looker treats each Explore as a unique entity.