This is an opinion post by Ewan MacLeod, originally posted on New Era Digital Partners’ LinkedIn Page and is republished with permission.
This week we held another Technology Counsel meeting with a senior executive from a bank here in the United Arab Emirates focusing on the management of data.
It’s a boring topic, yes.
Exceptionally boring, one might say.
But then again, every banker you meet tends to regularly use the phrase “data is the new oil“. Is it?
Data, is absolutely and completely NOT the new oil, at all, in most banks.
In most banks, data is the metaphorical equivalent of your children’s playroom when they’ve tipped all the Lego out everywhere across the floor. You know, thousands of bricks that, theoretically, would look magnificent if assembled in the right order. Nut no one has the manual, the Lego bits are all messed up, some are broken, 10% of the bricks are actually lost and no one can agree on where to start.
Or how. But someone needs to complete the Q1 update – so they’ll need to figure something out today.
Then once that report is filed, they’ll drop it and move on, until the next report is due.
It is not easy.
For example, another bank we worked with here in UAE does not know how many customers it has.
Or, to perhaps be a little more accurate, they can’t decide.
Hussein in the digital team calculates the metric this way.
Whilst Hamad in the CX team has a different query he runs. Samia in the Technology team has a different answer because she’s exporting directly from Core Banking.
Hussein is collating from a Core Banking CSV dump and using one of the product CRM tool outputs – and doing lots of VLOOKUPs in Excel. Whilst Hamad does it another way entirely — and crucially, doesn’t include dormant account users.
But when the figure goes to the board, ALL of them include dormant account customers… because that generates the largest figure.
Is this the new oil?
Not yet.
Not when there’s no single definition of what a customer ‘is’. For example, are we counting customer identifiers? Because often, you’ll find 5 unique customer identifiers. One in that system. One in this system. Another over here. And another because he’s got a joint account with someone else. And that system doesn’t allow duplicates. Oh, and it’s a totally different user account for the credit card system.
Have we got to the new oil yet?
No, because we haven’t even got an agreement on what we mean by how many customers we serve.
If there’s no Chief Data Officer or equivalent, or if there’s no team installed focused exclusively on data management, it’s often someone in the Technology team that universally and very quietly decided the definitions – because someone had to.
If that person in Technology has a bit of authority – a chief or EVP or similar, then hopefully those definitions can stick and bring some clarity.
We haven’t, you’ll notice, got to examining the actual data yet.
That’s where the Lego analogy gets really accurate, we think. Go and look at the customer records. Not many executives ever do. In fact, most executives have never, ever seen the raw customer data in their banking systems.
This is where we started, with our Technology Counsel meeting.
We talked a little bit about definitions of what ‘a customer’ means.
Again, most executives are just presented this data and it’s only if they ask the right questions will they be able to delve deeper.
We started by scrolling through some CSV dumps from the key banking systems – some Core, some CRM.
It’s amazing how many customer phone numbers are set to: 99999999 or formats like +971+97122332233.
Any customer record created before 2010 tends to have the equivalent of mickey@mouse.com in the email field. Which, unless you’ve upgraded the CRM recently, is stuck somewhere in an extra field that the Infosys guys had to manually create.
It’s only when you do a search for a@a.com email addresses and find there are 5,000 of them that you begin to realise there’s a data quality issue – because the branch network staff didn’t enter the right data during customer onboarding. But the field required something to be entered in order to proceed.
Or search for +971 (the country code for United Arab Emirates) and you find dozens of variations – everything from:
- +971 (correct)
- +9 71 (wrong)
- + 971 (wrong)
- +971971 (wrong again)
- 971 (wrong)
And so on. We’re labouring the point.
But if this is how your basic customer records are looking, how’s your “Oh, data is the new oil,” statement feeling? It might be. For some banks and some companies.
Garbage in, garbage out
So we started here. We put the CSV files up on the big screen and we scrolled through them. We looked at the actual naked customer data. Some of it was exactly as expected. Most of it was a total mess.
We made notes.
We examined.
We discussed.
We highlighted certain rows, searched for duplicates and so on. We found loads of incorrectly formatted email addresses.
No wonder customers keep hounding the call center, our executive mused.
No wonder the response rates are very low when you’re sending thousands of SMS messages to incorrectly formatted mobile numbers or out-of-date landlines.
This is where we started — to give the executive an appreciation of what data they actually had in the system.
Of course, you can’t just edit that customer record and correct the mobile number or email address – even though it’s clearly wrong. There is a process for this. There’s a methodology. There is (or should be) a clearly assigned data manager and owner for each set of data. The base-level customer records need to be cleaned and sanitised. This has to be done carefully and correctly.
In many cases, you can actually ask the customer to do it themselves. In many cases, the customer will be only too happy to correct incorrect data.
But there’s a process for this. There are compliance, regulatory and legal considerations to review first.
Before you can begin to move toward the New Era (see what we did there?) of data management, enrichment, utilisation and exploitation, you have to get the hygiene right.
We can’t start the data mining conversation yet. Nor can we even begin the machine learning or generative AI discussion.
But this is a start. Get the data cleansed. Know who your customers are. How many you’ve got and what the definitions are, or should be.
Then let’s look at the next step.
It’s exciting — really exciting — when you can begin to march forward purposefully in the field of data.
It really is empowering for everyone in the room – genuinely. Because you can build upon this, steadily and progressively and you can begin to see how the phrase ‘data is the new oil’ can actually mean something in a bank.
Thanks again to our customer who has given permission for us to publish these insights into topics we cover in our Technology Counsel sessions. You can find out more information over at our New Era Digital Partner services section – and if you think we might be able to work together, please drop me a note.
Finally, did you like the Midjourney generated ‘data is the new oil art’ at the top of the page? 🙂