The Business case for the AMAZON bank
This post is not should Amazon create a bank, just would BigTech companies do better at the retail FinTech/ banking functions because of the data they hold on you and me?
All of Amazon’s reporting data is here, Q4 will be released on 31.1.2019. We know from the existing reporting that they have passed 500,000 employees, added $5Bn to cash and sales income of over £200bn. Whilst the Q10 reporting give numbers, facts and a short section on risks there is no mention of data. Can AMAZON or any other BigTech buy an existing international retail bank in cash – yes, but would they be able to do any better?
Ignoring the specific relationship that customers have with Amazon and these same customers with their banks, the regulation frameworks and that Amazon has a range of credit/ gift card offers right now; the question:- “Is any of the BiGTech’s data set on me (and you) provide an advantage insomuch that it would enable them to make better decisions as a “bank” that say your bank?”
Why pose the question? Banks are really interested in the concept that access to more data will enable them to build new services, improve offers, uplift services and create new margin. However even if they had access to any new data (retail, social, health) would anything change? Let us therefore start with Amazon’s data and then look at some examples to see if they can do a better job.
What Data “could” Amazon have?
Verifiable name. From address and bank details they can verify your “name”
Verifiable addresses. These are places you have delivers to. These can also be cross checked with public data (electoral roll) to confirm who lives where and with whom.
Family. Both from content (children) and from name/ address data with different account names.
Friends. A percentage of Amazons core business is gifts and the delivery thereof. This provides a unique set of additional data on friends, their location, their spending power and other social insights.
Bank details. Probably that you use more than one bank account, credit card. Who the insurer is, how often this changes, decline frequency, debit or credit.
Time and frequency. How long have you been with Amazon and you purchase history
Returns. How often you return and why
Social media. Given the sharing capability, some data that links you to your social media activates
Search/ Intent. As you search for items tell what you would like and are thinking – this gives a profile of possibly both wish and desire.
Change (delta). The patterns of purchases, time of day, types of items, confidence.
Propensity to nudge. Customers who purchased this bought this, these items come together, if you liked this – you will like this. Your behaviour.
Disposable income. Probably not directly but from the items, spend and address the maths would be able to give a reasonable prediction.
Trust. How you look at reviews, how they influence you. Where you buy items from and types of retailers
Driver. lowest cost (price) vs quickest delivery (urgency)
Search intelligence. First item on list or recommendation or detailed search for the right product, right price and right delivery
Propensity to review. Content that you provide, the language and style. What you review and how.
Leader/ follower. From items how do they position you in the overall adoption curve
Voice (Alexia). How often you do things – this enriches that you purchased some content to how often
Adoption/ learning. The content consumed, the questions asked (voice and typed), the frequency
Stuff I cannot see. Cookie data and lots more deep IP address, and device data, what devices, how often they change, from which locations. The dark world or tracking.
Three examples worth thinking about
Z = α + β1X1 + β2X2 +….. + βnXn
The core product of any bank is the ability to provide loans based on BASEL 3 – matching of assets to liabilities as a capital ratio test, dull but important. Risk is a judgement based on data about the customer. There are a stack of statistical models and moving to AI models, but the all need data. Now if you’re a bank/ FinTech then you have access to banking type data, such as the example below. Everyone is the same, you just tweak the makeup.
AMAZON can base a credit score on a wholly different basis. Why is this interesting, as it changes the risk profile, which changes the margin and the appetite. The above model is very easy to manipulate, Amazon’s data is harder to manipulate unless you want to over a very long time.
AMAZON model will be a combination of their own data set, which only they have access to. They can easily do a linear regression model and then improve this over time to refine a very sophisticated model based on behaviour of purchases and not on default.
Who wins in the credit score world - AMAZON data is superior as it is far more suitable for AI/ Machine learning.
The current Fintech offers are vanilla (at best). One bank, like any gas, water, electricity, mobile phone SIM is a utility and differentiation is created by the complexity of the bundle to confuse the user and avoid direct comparison. Bank accounts are bank accounts and loans are loans. Access to, risk and time provide massive product variation.
Enter affordability. The current model is that if you want it now; we give you credit, go buy. (unless you cannot get credit) Credit means that the cost of your desire/ urgency has taken a premium price to an even higher overall price including finance charges, which significantly disadvantages parts of society. Amazon can offer something different based on their data.
AMAZON know you have searched for that TV, they know you can afford it and can give you credit but they can also improve their supply chain, guarantee future income to manufactures, offer you a better price and increase their Brand value with a subtle change. How. They offer to the user to put aside some income from (salary – costs) disposable income for several months, hitting the target immediately releases the product. This means that the price may be lower, demand is staged, it is more affordable (ethical). Yes the user has determined that they are prepared to wait.
Why interesting – the existing credit model is based on charges for the loan. This model is based on improving margin by helping with manufacturing inventory and logistics. It also taps into a more sustainable economic model and brand value. It taps into a one of the core differentiation data sets that Amazon has - propensity to nudge.
Given the salary, union and other negative effects of 500,000 workforce, this could help with a move to being more ethical with staff and customers. What a position!
Who wins in the affordably world – AMAZON as their infrastructure is superior as it is far more suitable for AI/ Machine learning support and insight.
AMAZON could say to its staff that their salary will be paid to the AMAZON bank account as from next month – 500,000 subs straight away.
Salaries at Amazon.com Inc range from an average of $58,461 to $147,184 a year. A Picker make the least with an average annual salary of $25,517. More on Average salaries in US, UK, Australia, India gives rise to a monthly deposit of say $15bn per month ($30k * 500,000) Assuming that 5% says in the account for month end, this builds a deposit base of $750m per month.
Why interesting – AMAZON may say to its staff that they will share the upside from using the aggerate money to lend to their retail customers or supply chain as a B2B, and give the other percentage of contribution to the AMAZON investment foundation that funds new startups using the AMAZON platform to build new services and new innovation. AMAZON benefits from the growth and share price increase, everyone else from access to a new upside. The Amazon bank could be a co-operative?
Who wins in the scale world – AMAZON as their ability to grow a self-serving eco-system is superior as it is far more suitable for AI/ Machine learning support and insight.
At a totally superficial level Amazon looks like it should become a bank, or it should start to provide the AI output to the FinTech industry and incumbent banks. Worth noting is that even if an individual takes their own data back (which I favour), the individual holds the data and not the algorithm to determine what services I should be offered. AI needs the amalgam of these large data sets and that puts BigTech and Communication/ media companies in a commanding position.
Now which corporate is going to make the most of thinking before the regulators get way to excited and change the market structure?