How transaction analysis works
We've built a whole page for this, here. In summary though, it works like this:
- Movem receives raw transactions.
- We look for income and rent, based on internal algorithms, rulesets, criteria and years of experience.
- We ask the applicant to confirm if we are correct. If we are, the income is shared with the client. If not, then a manual review is required.
The applicant cannot bypass our benchmarking and algorithms. Everything Movem verifies as income, has met all of our criteria.
This is a broad word for a process or set of rules to be followed in calculations and other problem-solving operations — typically by a computer.
Our competitive advantage
Firstly, let's get something out in the open: Movem is not the cheapest provider out there, and probably never will be. This isn't because we're greedy, it's because our product is conceptually different to our competitiors, and has more costs associated with it.
But how can we explain why our transactional analysis is so much better, without tearing down our black box and showing you the insides? Well, let's try. Oh, and it's also here. 🤓
It starts with how we receive data from our partners. We receive data in a format similar to this (like everyone else):
Amount, payment reference, date:
- £1000 | AMAZON GB 131421314 HB8779 | 11/11/18 - £56 | CURRYS 62487283 | 10/11/18 + £10 | MUM | 07/11/18 + £2000 | MOVEM | 01/11/18
Now, there's a process called data categorisation, or CAAS (categorisation as a service). What this does — broadly — is take a transaction and attempts to categorise it. For example:
AMAZON GB 131421314 HB8779 | TAG: SHOPPING CURRYS 62487283 | TAG: SHOPPING MUM | TAG: INCOME MOVEM | TAG: INCOME
This is what challenger banks, like Monzo, Revolut and Starling, do to provide more analysis of your spending. They will use some form of automation to tag transactions.
But look at the 'income' tag. At first glance it's worked, income has been tagged correctly. This is how they're 99.8% accurate — because they're tagging an incoming payment as 'income'.
But that's the easy part. If this is all you need, you don't need Movem. Go directly to an AISP and use their tagging.
Now consider you've looked at 18 months of transactions, and this is what your 'income' tagging looks like:
MOVEM | £1000 | 01/12/2018 | INCOME DELIVEROO | £1000 | 01/11/2018 | INCOME MOVEM | £1100 | 03/10/2018 | INCOME U1REFUND | £100 | 01/10/2018 | INCOME MOVEM | £1000 | 01/08/2018 | INCOME MOVEM | £1100 | 11/07/2018 | INCOME MUM | £100 | 01/07/2018 | INCOME MOVEM | £1000 | 01/06/2018 | INCOME MOVEM | £1000 | 22/05/2018 | INCOME PAYMENT1314 | £1400 | 01/04/2018 | INCOME PAYMENTHH | £1000 | 06/03/2018 | INCOME DELIVEROOUK | £1000 | 01/02/2018 | INCOME MOVEM | £900 | 01/01/2018 | INCOME GBEA | £1000 | 01/11/2017 | INCOME MOVEM | £900 | 03/10/2017 | INCOME 4244HHHnb2 | £1000 | 01/09/2017 | INCOME MOVEM | £1000 | 01/08/2017 | INCOME DELIVEROO | £1000 | 01/07/2017 | INCOME MOVEM | £1000 | 01/06/2017 | INCOME MOVEM | £1000 | 01/05/2017 | INCOME
Now what? Do they have a regular income or not? They clearly have income, but is it regular? Can we rely on it? How many jobs do they have? Some months are missing, was this a holiday or did they lose their job?
It looks like they have a few different sources of income... are any of these consistent? Do we think they'll continue?
How do we make sure none of these payments are from friends and family? Or one-offs? What about bonuses?
It gets very messy very quickly. Believe us — anyone that says this isn't a major issue, has no idea how deep the rabbit hole goes. 🐰🕳
So, Movem is an additional layer to data tagging. We call it [data contextualisation](#Data contextualisation).
The purpose of this is to take transactions that look like income, and ensure they meet out internal benchmarks for what 'regular income' is.
Instead of showing:
£1000 | MOVEM 7778HHSBNF DEC | 11/12/18 £1000 | MOVEM PP8898JKSJ NOV | 12/11/18 £1000 | MOVEM J&£BR*£NND OCT | 11/10/18 £1000 | MOVEM 77JDS&&@^D SEP | 10/09/18 £1000 | MOVEM REFKSHHN££ AUG | 11/08/18
Employment: Has worked at Movem since August 2018, and earns £12,600* p/a.
*We see take home pay, so backwards-calculate it to a gross salary.
Contextualising the data is about taking a bunch of information that makes little sense, and converting it into a meaningful analysis that can help us make a referencing decision instantly.
Here's an example of just 5 of the questions we're answering with our data engine:
- Do all these payments look like they come from the same income source?
- Are the payments regular (both value and frequency)?
- Are there missing payments? Why? Are they typical for that kind of applicant?
- How confident are we that they'll be paid again, and when?
- Do these payments look fraudulent?
This has all been built in-house, and no other company has access to our engine.
We call it our unfair advantage.
Re-accessing an applicant's account
Movem cannot, and will never be able to, access an applicant's account after the reference is complete. We're given a one-time snapshot of their transactional history, and that's it.
Visibility of login credentials
No, Movem can never see your login credentials. These are handled by another highly FCA regulated company — an AISP.
We cannot see keystrokes, or any tracking whilst you're logging into your online banking.
Even in a worst-case scenario, where Movem was compromised, your banking credentials would not be retrievable, by anyone.
This is a type of regulation specific to companies that connect with banks. It stands for Account Information Service Providers.
Verifying new employment
Here's a good rule of thumb: to instantly verify income, we need enough transactional data to be sure that you're actually employed, and not just getting money for your birthday.
Typically this would be 3 months of data, however it can vary depending on frequency of pay, time of year, amount of pay, job type and many more factors.
So yes, we can verify new roles — but not always instantly.
When using the hybrid — our flagship product that uses humans whenever required — we will manually check any role that wasn't verified instantly.
Full visibility of transactions
Only the transactions that you've confirmed are either income or rent, and consented to be shared, will be visible to the client.
Movem can verify self-employed income in exactly the same way as regular employment. However, they are unique algorithms and rulesets, which may be stricter than a more traditional income.
Through our AISP partners we can verify income from foreign banks, in almost any currency in the world. It will be converted to GBP based on the current exchange rate.
Social and housing benefits
Movem can be used to verify incoming benefit payments, however it will not always be a suitable replacement for verified employment. View our referencing criteria for more.
Gig economy and on-demand work
Movem is great at verifying less traditional income, such as Uber, Deliveroo and Lift. Applicants will go through the same form, and their income will be handled using different algorithms and rulesets.
Verifying mortgage payments
Occasionally applicants will be making regular mortgage payments, but due to a change of circumstance, are now looking to rent. Movem will verify previous mortgage payments in exactly the same way as rental payments.
Using transactions for profiling
From day 1 we've been clear that we do not profile our applicants, or allow our clients to use bank account data to profile their tenants.
We've invested in internal tools to reduce the possibility of profiling.
As an example, we may strip back the payment reference if we believe it's related to a political party.
£1000 | LABOUR PARTY INCOME 4444233 | 11/11/18
Would be shown to our clients as:
£1000 | INCOME 4444233 | 11/11/18
Differentiating bonuses from salary
In 2018, we launched our 12 month analysis, which just verified 'income'. However this was always just the first stage of our greater long-term plan.
Right now, Movem can detect bonuses, commision, underpayments and more, however it doesn't catch 100% of the additional payments. Let's be honest; it probably never will.
But — and this is a Movem promise — you can guarntee we won't stop until we're as close as possible. Over time we'll be providing richer analysis and contextualisation than ever before.