I have a site that contains the banking information and a custom number (order number or billing number) that people must put in the wire transfer. It's generated like that XYZ, where X = some random number, Y = zero-padded userid, Z = increasing number for that account. If a user visits this page, a new order is created in the background with status "unprocessed", or the same order number is used again, if it wasn't "completed" by an incoming wire transfer before.

At first, I used an online banking tool where I copy&pasted all incoming wire transfers into my web application, which would regex for all ordernumbers, do a little plausibility check (incoming amount equals the amount that was set when the order was generated; user has open orders) and set the order to "processed" and extend the account by the according amount of months.

Now, I use a VPS where I installed an HBCI client (AqBanking). It's relatively easy to set up and once a day, it fetches all new wire transfers via cronjob and sends them via cURL to my web app where the same regex and plausibility check happens that I used before.

It feels a bit weird to store your banking PIN on a VPS, so make sure the VPS is properly secured.

The VPS sends me an email whether or not everything worked and that's it.

Most days of the week, there's no problem. Sometimes, you get the wrong amount from people, some people don't put the order number, some people confuse the bank number (BLZ) with the order number and so on. In these cases, the regex parser fails and tells the reason in the email. I then process it manually. Some people think that wire transfers work instantly or that they are also processed by banks over the weekend (which is not the case). I have a few email templates to answer these requests if necessary. Fortunately, it's not that much effort and most of the time, there's no manual work to do.

I didn't adjust the system for the coming IBAN stuff yet. Have to look into that.