Toshl Finance Blog

Spending Categories That Learn from You

When an expense or income flies in from a connected bank account, it’s in a raw form. We first need to get to know it better and make it easier to understand at first glance. In most cases, the data that’s available from the bank is the amount, date, financial account and a description of the transaction. This description can be verbose and include info about a specific location, details of the vendor. More often than not, it can also contain just a few cryptic numbers or text that say very little to the untrained eye.

To make this information more useful for analysing your finances, the expenses and incomes need to be analysed and the true meaning of it determined. This means automatically attaching a category and tags, often based on a pretty flimsy description. This process can then tell that a scant point-of-sale description like “LIDL-SI 227” really means category: “Food & Drinks“, tag: “groceries“.

Initial categorisation

The first step in this categorisation is usually done by our bank connection partners like Salt Edge and Plaid, which use anonymous aggregate analysis of millions of transactions to better determine the likely meaning and purpose of this purchase or income.

This however applies general rules that are valid for most people. If we take the example from above, most people will go to Lidl to buy groceries. However, your case may be different. What if you categorise your expenses with custom categories and tags, or visit that shop for a different purpose than most? What if you only visit Lidl to buy cleaning supplies for your racoon farm pet project? You’d edit the expense to re-categorise it as e.g. category; Projects, tags; racoons, cleaning supplies. Doing this editing every time you shop there would be quite a nuisance. This is where Toshl’s personalised categorisation comes in.

Personalised categorisation

In addition to the initial categorisation, Toshl will track your corrections of the categories and tags and match those changes to particular elements of the description. In other words, Toshl will learn from your past choices and apply it to new expenses as they get categorised. Future expenses with similar descriptions will get the categories and tags you chose for other expenses that had those kind of descriptions.

Your racoons will get the recognition they deserve. Finally.


Constantly improving

No automatic categorisation system with such a variety of data and choices can be perfect. Sometimes the available raw data just doesn’t suffice to make an accurate determination, sometimes the use case is just so specific it would be impossible to guess. We’re however doing our utmost to improve and bring you the correct categorisation in the vast majority of cases.

The more you use Toshl and adjust your categories, the more precise it will become. This adjustment feedback can be incorporated both in the initial categorisation as well as the personalised one. With initial categorisation particular venues can get their default categories adjusted, as lots of people anonymously correct the category for a single venue. Your category vote here however is weighed against all others.

With personalised categorisation, the learning will be done specifically for your user account and only your data, so the changes in automatic categorisation and tagging will happen much quicker.

Connect your bank or credit card with Toshl

The automatic categorisation is available with the automatic bank connections in the Toshl Finance apps. Toshl Finance can connect to more than 13 000 types of financial accounts all around the world. Try out the free 30-day trial of Toshl Medici.

Toshl monsters are already learning all possible categories by heart. Just don’t break their hearts by whaling.
Posted in Uncategorized