Order extension for bulk discounts

[jace]

As a general principle, it is always better to take the customer's money first and let them deliberate on options later. We practice this principle in ticket booking when we ask for attendee details later. One area we could do this better is with bulk discounts.

Assumption: When a buyer is unsure of how many are attending, they'll hold off on placing the order, instead of giving us cash and cancelling later.

Solution: Allow a buyer to extend the order from the ticket assignment page. While new purchases will constitute a new order, orders can have a "parent order" column that links them together (if the parent also has a parent, the chain is traversed to find the top-most). When calculating discounts, line items from across all orders are used. Older orders never get a cashback. Discounts are always applied to the new order.

UI: The ticket assignment page can have the widget embedded for additional purchases. Since the ticket assignment page can be accessed by anyone with the link, a person can forward this link to a friend/colleague for the additional purchases. This also introduces the risk that their own ticket can be cancelled/reassigned, so this needs to be considered a little more carefully.

[jace]

This assumption needs validation since a counter assumption can be true as well: by only offering bulk discounts in a single order, we encourage buyers to commit seats upfront and cancel if there are no takers. This needs a test.

Assumption 1: Buyer abandons order because they can't find takers for a bulk purchase. We don't know how to distinguish this signal from every other reason to abandon cart, including sticker shock or "just browsing".

Assumption 2: Buyer purchases in bulk and later cancels because there are no takers. This one can be distinguished from other reasons to cancel if the ticket is unassigned at the time of cancellation.

Since these assumptions aren't mutually exclusive, having data for only one is insufficient. We need a measurable test for the first assumption.