Use machine learning to estimate how long a processing transaction will take?


(i know the first 3 numbers) #1

An example being an amazon refund or if you’ve just sent a payment out and it shows as pending, having an eta from fed information would be a really cool idea


(Alex Sherwood) #2

That’s a cool idea :heart_eyes:

Amazon might not be the best example because they take your payment once your order is ready to be shipped & obviously that can either be instant or a week, depending on what the item is & where it’s being sent from.
Transactions from TFL & other merchants that delay their payments would probably be useful though. And knowing how long it typically takes a merchant to present the payment (so when it’s likely to actually be taken from your account & not just your balance) would too.

My only concern would be users contacting the support team because a payment’s taken longer than the estimated time, to appear. But hopefully Monzo could think of a way to manage that.


(Nigel Ng) #3

Hey @HitechCasio,

Nigel from the data team here. That’s a really good idea! We probably don’t need machine learning for that, but just summary statistics for presentment times for each merchant (think histograms of time to presentment). Maybe in the app we could say something like “90% of transactions with this merchant settle in x days or less”. I’ll talk to our internal product team to see if we can get this into the app!


(Alex Sherwood) #4

A post was merged into an existing topic: Monzo Job Posts :loudspeaker: