Data warehouse and BI
HubSpot is a source. The warehouse is the join.
Snowflake, BigQuery, Power BI, and Looker should read HubSpot on a schedule you can explain. They should not write back into properties unless a named job owns that write.
| Job | Direction | Owner |
|---|---|---|
| Extract | HubSpot to the warehouse, on a schedule | The person who can say how old the extract is |
| Join | Inside the warehouse, with billing or ERP | The analyst. Not a contact property. |
| Writeback | Warehouse to one named HubSpot field | A job with a name, not a dashboard on a timer |
| A board | Reads the warehouse | Whoever will explain a number that disagrees with the CRM |
A dashboard that updates a HubSpot property every hour will fight the rep who just edited it. Direction is part of the design.
Say how old the number is allowed to be.
A board that argues with this morning’s CRM is usually a stale extract, not a new insight.
See sync direction