Agreed, and this could allow people to have a self funded overdraft of sorts where they have a limit equal to their savings (i.e. standard monzo pots). Go beyond that limit and then the overdraft charges should apply as that’s when they are spending monzo’s money.
Anyway like you I’m not holding out hope. As much as I’d like this approach I don’t need it, and I’m happy enough with the rest of monzo’s approach that it’s not a deal breaker for me.
Ditto really. But I think a change here could really help. When I was younger and used to live in my overdraft, trying to manage myself out of it each month was a challenge.
I think if Monzo let pots count towards your balance, then lets say you’re £500 overdrawn and you get paid £1000, you usually spend all of that money, but you want to pay back £100 this month so at the end you’re only £400 overdrawn. It’s easy to say “just don’t spend it” but at the start of the month if you could move £100 into an “Pay Back Overdraft” Pot, then it’s half the battle I think. Out of sight, out of mind is a big reason a lot of us use pots.
True. It’s probably a bit flippant of me to say that I don’t need it and therefore it’s not a deal breaker, it’s just this has been discussed to death on here and I don’t see it changing any time soon if at all.
I do see it as a really valuable way to do things for those that would benefit from it the most. I was definitely in that camp a decade ago and it would have allowed me to feel even just a little more in control of my finances and would have reduced the stress I felt at the time.
I use this very loosely, but it’s a bit “Helping the rich get richer”
If you’re lucky enough to have a bigger balance and you keep upto £2k in your account, you can gain interest on all of it, no matter which pot.
But if you’re struggling and you want to try and save £5 a week to pay back your overdraft, you can’t gain the extra help that pots would bring.
What’s your elevator pitch - what are you going to differently compared to the other banks out there?
If only it was that easy 
I’ll split up credit limit and price:
Credit limit decreases
We have two processes here
Credit limit increases
We don’t do this proactively and customers need to “apply” for a limit increase - you can do this any time in the app by going to “manage your overdraft limit” (or something like that).
There’s a few reasons for not doing it proactively.
First, this comes with more risk since we need to make more assumptions about your current financial situation, eg. we can’t ask you if your income has changed before giving you the limit increase.
Second (and the main reason for me) is that it can be quite predatory if not don’t with a lot of care. You’ll tend to give a limit increase if you think you can make more money by doing so - which means you’re hoping that a customer using eg. £900 of a £1,000 limit will start using £1,800 of a £2,000 limit which may not be a good customer outcome.
Some people don’t have a great relationship with credit whereby if they suddenly see their limit has been increased from £1,000 to £2,000 then they’ll go use the additional money even if they don’t need to. I’m sure you’ve heard stories of this kind of thing from friends/ family (I know I have).
That’s not to say we won’t do proactive limit increases in the future but we want to take care in doing so and for now there are other things we’d rather get right.
Price changes
We don’t do this either at the moment. Personally, it’s something I’m really keen on but just hasn’t been a priority yet.
To see if I could get a better rate, would I have to gamble by disabling my overdraft and re-apply hoping that I get offered no worse than what I have currently? Or is there a way for CS to check what I’d likely qualify for?
It’s not massively relevant yet.
In borrowing you need to gather data for about 6 months - 12 months until you can get a “read” on whether to change decisioning in a meaningful way.
We have no manual review so we don’t manually review if you select “other”.
Exactly - it’s still important or us to know for other reasons!
That’s interesting thank you. I know it’s unlikely that you’ll be able to answer this but I guess selecting ‘other’ has different weighting when compared to all the others because it’s so broad?
You could literally be funding someone to start a drug empire or someone like me wanting to start a competitor bank 
If they were honest of course.
Just to be clear, I was actually looking for a loan the other month that didn’t really fit into any of the other categories and it was for neither of these reasons 
So I was wondering if I tried to get it to fit into one of the exiting categories, would it offer me a better rate.
Finally some easier questions 
What I do, the decisions I make, the products we build in borrowing really impact peoples lives. It’s no secret that things at Monzo mid - late last year weren’t so hot and I thought to myself - where else would I go? And the options I thought of fell into three camps:
I hate it! Last March it was ok for a month or two when it was sunny. But even though I have a commute of over an hour and at Monzo it’s optional to come in now - As soon as the office opened up about 2 months ago I started coming back in 4 days a week. We have a balcony at the “new” office as well which is :chef-kiss:.
Monzo’s only been around for ~5 years is can I just say everything will be twice as good? On a serious note - my goal is that we’ve revolutionised borrowing which has a knock on impact on the older high street banks - in the same way that I think challenger banks (Monzo, Starling, etc.) have changed current account banking and although you can argue the street banks may be catching up - they’ve been kicked into gear and been forced to catch up. This hasn’t happened in borrowing yet but it’s hopefully coming 
Afraid not.
Note to self: Add in rule to decline Ordog for any loans or credit pots.
Out of interest what was it? We can always add categories - if it was drugs or the bank of Ordog (or you don’t want to say) then feel free to DM me.
What does your typical work day look like?
What tools do you use to do your job?
Sorry, not going to give my full opinion on this (in relation to potential future products) in case it gets taken out of context.
FWIW - I’ve given my opinion on the “classic” credit card before though.
But who knows - this is just my opinion and i’m just one person in a team of ~80 in borrowing.
I think these are still “interactions” - maybe “interactions” was the wrong word though.
Some other things we can look at: are people missing payments, if they are do they catch them up, do they contact customer service to do so, are they repeatedly missing payments, are people “using” their overdraft, how much are they using it, why are they using it, what are they using it for, if they don’t already have one - are they applying for loans/ overdraft, are people changing their credit limit a lot.
From an overall UK population perspective - Unfortunately I think it’ll look fairly similar to today. It’s quite hard to really break into lending because there’s a lot more regulation, learning and data gathering required compared to building a current account.
That said - I do think the types of things you mentioned (Tymit, Starling, Monzo , etc.) will have started to make real dents in unsecured lending (credit cards, loans and overdrafts) in much the same way that today we’ve made dents in the current account market.
Making things completely seemless behind the scenes. You don’t see it but we take every opportunity to remove complexity from customers - two examples are we don’t have any manual decisioning for loan/ overdraft applications and it’s not perfect but it’s seemless when it works (and we’ll get it working better and better over time). Another place is detecting if you’re getting frauded or scammed when making a transaction and adding in friction only when we believe it’s necessary.
I did physics at uni and as part of that did some coding so I knew I wanted to go into a job that did some kind of coding. I then accepted the first job that took me which just happened to be a credit data analyst at RateSetter (a P2P lender). Turned out I really enjoyed it and joined Monzo ~2.5 years ago because I also love joining earlier enough you can have a large impact.
Thanks for answering! I often see on the periphery of all things “big data” python scripts to pull data / do analysis / other scripty things which I definitely understand.
Is that kind of coding for the actual number crunching / data vis what you do in your day job? (My day job stops at excel and powerpoint sadly).
Ooh follow up question - have you got any good examples of massive data sets you’ve used to help make one of those things seamless for the user? Would love to hear stories like that 
Would love to know the answer to this too!
It isn’t at all clear from the manage / initial change your limit screens if going through a couple more clicks could actually land you worse off than you currently are.
How long until credit pots 
Ohhhh - my favourite topic and I can answer a whole bunch in one go - this may be a long one.
The thing I’d take with me is Monzo’s data infrastructure setup - for someone who loves data it’s honestly unreal!
Monzo runs on AWS and uses a key value pair database called Cassandra. This is a NoSQL database and sucks for performing analytics. So almost any time anything happens “an event” is emitted with all the information you could want about that event. That could be you taking out a loan with data about the loan_id, interest rate, loan value, repayment date, it could be making a transaction with the transaction_id, payment scheme, amount etc. or it could be opening the app, sending a message in chat, the list goes on. All events will have a timestamp and the user that performed the action. These events are then stored in Google Cloud Platform.
So we effectively have a complete list of everything that has happened in the past - but it’s unstructured - ie. you can count how many loans we have (by doing select count(*) from loan_created_event) don’t but you can’t see what has happened to those loans over time and perform much detailed analysis. So part of my job in the data team is to “stitch” these events together into a structured “data model” (or table) that’s useful for performing analysis. We do this using SQL (BigQuery) and dbt engineering workflow.
As an example, one of those core models is loan_stats - this has the position of every loan Monzo has ever done and the position of those loans (balance, arrears status, etc.) at the end of every day since it was created. I checked and there are about 50 events (such as loan_created, loan_schedule_updated, loan_payment_made) that all feed into this and about 40 SQL queries that run one after the other (dbt does this scheduling for us) to create loan_stats.
In terms of size, another example of two of our core models are:
I really like this video about what is big data - TLDR; It’s anything that can’t be done on just one computer. BigQuery is able to handle this kind of data because Google just throws more and more computers at the problem.
In total I think we now have > 4,000 data models (individual SQL scripts) now and they all run every morning between midnight and ~10am so you have updated data for the previous day available every morning. Whats more if you want to make a change or add a column - that will have run and backfilled itself by tomorrow.
So we now have nice easy to use datasets that aggregate everything up. We then use looker as our data visualisation tool which most people in the company have access to to perform any analysis they want from the datasets the data team produce. We also use ad hoc SQL queries for ad hoc analysis and google sheets for some things where looker isn’t so great (more complex tasks usually). For building statistical models we then use python mainly - the ML team will spend ~20% of their time here but I don’t work with models too much usually.
In terms of my time I’d say I spend ~20% of my time on the data model piece, ~15% of my time doing analysis in looker or google sheets/ SQL, ~15% of my time writing proposals (based on the analysis) and then ~50% of my time in meetings or discussing product changes. Building products really is a team effort which everyone (data, design, user research, engineering, etc.) feeds into. These tend to come in phases so one week I might only be doing data models then the next might be writing a proposal
Sorry if that was too detailed - I should probably do a talk on this stuff once things open up!
