It is a fact table, the benefit I'm highlighting here is two-fold:

1. I didn't need to recreate my dimension tables, I simply reused the satellite table's temporal surrogate keys that I never need to manage. The resultant SQL join meets the right-deep join tree pattern star-joins are designed for.

2. transactions should be modelled as link-satellites with a dependent-child key defined as either the transaction date or transaction id in that satellite table. With the bridge table I added calculated outputs.

Hope that helps