In short, both the Paid orders export and the "GET /reports/paid_orders" API call returns null on the "shipping_tracking_number" field for orders, even if they have been shipped and even if "shipped_at" is not null. Possibly coincidentally, the example response on the API has both as null.

For clarification and more detail as to why this is important to me:

We've recently started to do 0.01 shipping and handling on ALL items so I'm monitoring very closely the "net' of each item (total amount paid per item minus actual shipping costs). Since more than one item can be on an order I divide the label cost by the amount of items on the order. However, since shipping costs aren't [directly] calculated, customers do not always combine their items into one order, so we catch it on our end and may ship multiple orders in the same box and manually update the other orders with the same tracking number. So, while there is tracking data in the shipments API call it can only be cross referenced with the one order the shipment was generated on.

I'm sure that's a unique use case, however there are plenty of other reasons this could be beneficial to others if fixed.

More Information & Testing:

I pulled all the data YTD with both the Paid orders export and through the API and noticed there was actually one order with information in the tracking field. A manual entry for a pickup we cleared off this morning. To double check this I manually entered a tracking number on a different order and sure enough it showed up in the data. Since we do manual tracking entries almost every day I can assume that this data is just disappearing at the end of each day. However, if you go to the order page of any older item (in my check, an item shipped on 1/3/24) there is tracking on the order page in the shipment section and the order tracking section. So, the data is SOMEWHERE. I'm sure it's just getting overwritten with null at the end of each day, for manual tracking entries at the very least. Why generated tracking numbers at the time of purchase never make it into this data I'm uncertain. Seems like it might be an easy fix.