I'm designing the 2nd major iteration of a relational database for a franchise's CRM (with lots of refactoring) and I need help on the best database design practices for storing job invoices and invoice lines with a strong audit trail of any changes made to each invoice.

Current schema

Invoices Table

InvoiceId (int) // Primary key
JobId (int)
StatusId (tinyint) // Pending, Paid or Deleted
UserId (int) // auditing user
Reference (nvarchar(256)) // unique natural string key with invoice number
Date (datetime)
Comments (nvarchar(MAX))

InvoiceLines Table

LineId (int) // Primary key
InvoiceId (int) // related to Invoices above
Quantity (decimal(9,4))
Title (nvarchar(512))
Comment (nvarchar(512))
UnitPrice (smallmoney)

Revision schema

InvoiceRevisions Table

RevisionId (int) // Primary key
InvoiceId (int)
JobId (int)
StatusId (tinyint) // Pending, Paid or Deleted
UserId (int) // auditing user
Reference (nvarchar(256)) // unique natural string key with invoice number
Date (datetime)
Total (smallmoney)

Schema design considerations

1. Is it sensible to store an invoice's Paid or Pending status?

All payments received for an invoice are stored in a Payments table (eg. Cash, Credit Card, Cheque, Bank Deposit). Is it meaningful to store a "Paid" status in the Invoices table if all the income related to a given job's invoices can be inferred from the Payments table?

2. How to keep track of invoice line item revisions?

I can track revisions to an invoice by storing status changes along with the invoice total and the auditing user in an invoice revision table (see InvoiceRevisions above), but keeping track of an invoice line revision table feels hard to maintain. Thoughts?

3. Tax

How should I incorporate sales tax (or 14% VAT in SA) when storing invoice data?