____ Science Without ____ Context

The right metaphor for data is not that it's the new oil, at least not for banks. 

It’s fracking– slang for hydraulic fracturing, a process where a cocktail of chemicals are injected into rock cracks causing pressure to break open the surrounding shale, releasing oil and gas. 

In banking, that cocktail of chemicals is downward pressure on tech from executive teams who need actionable insights. 

The rocks are 1) systems of record that aren’t always connected and 2) BAU processes, many of which live outside tech’s official systems as semi-automated processes and user-defined tools.  So the completely legit ask for insight is never as simple as “go get the data from systems x, y, and z” because the rest of the alphabet is needed and missing.

And the value extracted— the tiny bits of oil or gas— is insight… but small and inconsequential because the data only casts light on a small part of a very large, dark cavern that can only be lit by data sources you simply don’t have (yet). 

Oh, and if you follow the anti-fracking community, you know about tap water catching fire when fracking occurs nearby.  So the final extension of the analogy– flaming tap water– signals the unintended consequences of making the wrong investments– the pressure that’s brought to bear on tech to spend time extracting insight from x, y, and z when they should be digitizing the rest of the alphabet.

[Alright, enough metaphors.  Let’s drill!]

Hard Truth #1

Banks spend a lot on data ecosystems, not what feeds them.

No one can complain about the amount of money banks invest in data infrastructure (billions) but, oddly, they do it without commensurate investments in business process redesign and digital workflow at scale.  

Why? In part because infrastructure is explicitly tech’s job while process redesign is– arguably– the business’s. Business folks are so hyper-focused on selling and servicing clients— maintaining and deepening client relationships— that they honestly don’t have the time for hands-on, continuous process redesign.  So the usual, well-intentioned business response when complex data problems bubble up to leadership is to assign whoever on their side is free at the time.  And that usually means a generalist, because there are never enough specialists. Ever.   And experience shows that neither they nor their tech partners successfully deliver process redesign because it inevitably demands business depth.

So neither side is faultless.  They both know that data is a worthwhile investment. And together they prioritize what everyone else in the industry prioritizes– the hydration of reference data systems and the creation of lakes upon lakes upon lakes.  But money is finite so the “masters and lakes” strategy means that banks struggle to invest outside of that explicit ecosystem, failing to create a continuous, end-to-end flow of the useful raw material of insight.

Insight #1: Before any bank can call themselves data-savvy, they need to acknowledge that their systems of records don’t yet store the right data because their existing business processes aren’t yet properly digitized.  In other words, before any can call themselves data-savvy, they first need to be process-savvy: put their operational BAU under the digital microscope— and invest *heavily* in digital workflows that connect their processes end-to-end, intentionally remodeling them to produce data. 

All banks use this kind of language but given competing priorities, struggle to put their money where their data begins– to seed the clouds of rain that feed the lakes.

Hard Truth #2

Banks rely on technical specialists— data scientists— who don’t understand the businesses they serve and worse, lack the business imagination that could allow them to reshape the future of those businesses.  The best statisticians (because that’s what data scientists were called before they were cleverly rebranded by big data vendors and banks got all excited about the new shine) learn what they can from generic new-employee training (i.e., some guy from the 1970s talking about the trade life-cycle)... but show me a bank that invests in continuous business training (outside of soft-skill leadership training for “high-impact” employees who have already demonstrated they have soft skills and don’t need training) and I’ll show you a corporate booster who has bought into an HR narrative that even HR knows is an underinvested sham… because HR is almost exclusively a risk function, sadly.

[deep inhale… deep exhale… refocus… continue.]

The path to data-driven insight— as a repeatable process— demands business context and depth. And unfortunately, the market for data scientists has been so hot since 2012 (when HBR played its usual role… of substance… and declared data science 'the sexiest job of the 21st century') that banks are satisfied to land specialists who are inadequately prepared to interpret our specific business context.  

They know statistics for damn sure.  But ask them about credit or collateral or securities processing and they’ll tell you what they’ve been told by the requirements documents they’ve read.  

It’s an institutional failing. Data scientists have no real learning path but to become reliant on the usual domain experts— banking’s business analysts— who themselves struggle with the art of the (business) possible.

Insight #2: It’s not just a waste of an expensive data specialist’s valuable time. It’s that they’ve been neutered of their data imagination.

Hard Truth #3

When business leaders finally get serious about data– once they hire data leaders with business depth (on both sides of the functional wall)– those folks tend to become token support functions.  Like Chief Diversity Officers, they don’t head businesses, so they don’t have real power (other than the power of influence), but their hiring and their explicit role absolve business leaders above them from hands-on ownership of the problem.  

Insight #3: Change is needed at the top but not the kind they think.

So What’s an Org to Do?

Hard Answer #1: Redirect more of the investment dollars that go into masters-and-lakes toward digital workflow– automating the processes that consume your masters and feed your lakes.  Consider it a virtuous cycle because digital workflows create a practical (vs theoretical) demand for masters and lakes.  

While you’re at it, you’ll also want to reimagine how you automate Ops processes.  To steal a line from that linked piece, you’ll need an intentionally-iterative mindset:  “If you’re designing any 5-step workflow, add a 6th step that only kicks in once a week/month/quarter that asks how the process should change.  If your first implementation of that workflow doesn’t automate all of the steps– i.e.,  it adds other value like much-needed structure and controls– then have your process factory keep coming back to it weekly/monthly/quarterly to see if your other tech tools are now ready to automate a step.”

Hard Answer #2: hire the kind of business-savvy data scientist who will force your non-scientists to stop using the language of AI/ML like it’s magic.  1) It’s just statistics. And 2) training those statistical models takes as much work– and ongoing maintenance– from the business– not tech– as the manual pain that you think you’re automating yourself out of.

Hard Answer #3: Demand that business owners own their data problem.  Personally.  Back in Henry Ford’s day, I doubt he had an Engine Executive, Tire Executive, or Cup Holder Executive.  

The point?  Aside from deepening relationships and building client trust, data is *the* business.  

And there aren’t many business leaders who actually understand what data is or how to use it.  

Delegating that responsibility is as corrosive as delegating the seniormost relationship management role.  

It’s poor leadership. 

–

So if you lead a business, take the time to educate yourself.  Or pay someone to educate you… but not to do it for you.

Maybe take that internal course on data management.  

And then fire whoever wrote it.

[That Insight Lightbulb Flickers to Life]

It should concern us when leaders delegate the important stuff.

It should concern us when functional expertise comes without business depth.

And it should concern us when banks redesign processes to require fewer people, as opposed to rearchitecting them to produce data-centric, data-generating, digital processes– the raw material of insight.  

These drills– of continuously investing in data infrastructure without commensurate investing in digital workflows– and I mean outside of the masters that they consume and lakes that they feed– are fracking pointless.