Problem auto-categorising Paypal, Curve and Asda transactions

A big issue I’m having with this app is it seems that only the first word of the description is used to identify similar transactions. For most places this doesn’t really matter. McDonald’s Bristol and McDonald’s Swindon for example would both correctly become “Eating Out”.

For Paypal it’s a bit different, all their transactions are “Paypal * ShopName” and Nova’s categorise similar transactions feature then attempts to categorise ALL Paypal transactions. Curve has a similar isssue because they use “Crv * ShopName”. Whatever comes after the * could be a very different category.

Asda has a similar issue, their transactions are labelled either “Asda stores [location]” or “Asda filling station [location]”, two very different categories, but again Nova attempts to categorise them the same.

I really don’t want to have to categorise every Paypal transaction manually, especially as I have a few shops I regularly buy from, and I certainly don’t want to have to do it for Asda. Is there a way to make more specific rules for these transactions?

1 Like

Hi Lum, Nova indeed tries to identify the merchant name or most important keyword from the description. So all McDonald’s transactions would be categorised “Eating out”.

Paypal transactions are harder because although they’re usually for shopping, they can also be transfers or anything else. If you had the possibility to create custom rules, how would you classify Paypal transactions?

I’d just do it based on each individual merchant, so “paypal * just eat” would be Eating Out etc.
Like I could probably do it with regexes but I doubt you’d want to do that in this app as it’s not exactly friendly for most folk, maybe an option in the “categorise similar transactions” screen that triggers nova to look for the second most important keyword (which you could then drill into and trigger looking for the third if you really really needed to. Internally I guess you could implement this by chopping off the shared keyword (paypal) and then chucking it into the same algorithm you used the first time

This method would also work for the Asda problem, since the next keyword would be either “stores” or “filling”