Decision Engineering™ in Practice Advisory engagements for financial institutions and healthcare systems.

Twenty-five years inside banks and health systems. Different geographies, different mandates - For 25 years, I’ve worked inside banks and healthcare systems across different countries and institutions. Different names, different technology, different leadership teams. But the same problem keeps showing up.

Most failures don’t happen because people had bad intentions or because there weren’t enough policies. They happen because somewhere between the boardroom decision and the actual day-to-day execution, things drift apart. And by the time anyone notices, the problem has usually been building quietly for years.

That’s the work I do.

I help banks and healthcare institutions find where that drift started, why it happened, and how to fix it before it turns into a regulatory issue, financial loss, or public failure.

What I actually look for

I’m usually not looking for “technology problems” or “policy gaps.” Those are often just symptoms.

I’m looking for the moment where what leadership intended became different from what the system or teams were actually doing.

A simple example.

A bank introduces AI for loan approvals. Six months later, all the numbers look fantastic. Faster approvals. More customers. Higher conversion rates.

But bad loans suddenly jump from 640 cases to over 2,500.

The AI model technically worked exactly as designed. The problem was that nobody clearly defined what the system should avoid doing. The AI optimised for speed and growth because that’s what it was told to optimise for.

That’s not really a technology failure. It’s a decision-making failure.

And the warning signs are usually there long before the damage appears.

The same customer getting different answers depending on whether they use the app or visit a branch. Temporary exceptions slowly becoming “normal process.” AI recommendations nobody can fully explain end-to-end. Rising manual overrides. Leadership asking questions the governance teams suddenly can’t answer.

When these patterns appear, drift has already started.

The framework I use

Over the years, I built something called the Decision Integrity Chain™.

The idea is simple.

Every important decision travels through multiple layers before it becomes reality:

Purpose → Strategy → Intent → Rules → Judgment → Decision → Outcome → Feedback

Most institutions focus heavily on the first few layers. That’s where the presentations, policies, and strategy documents live.

But problems usually happen later, where systems, rules, people, channels, and real-world execution interact.

That’s where decisions slowly change shape without anyone fully noticing.

The framework helps organisations trace how a decision actually moved from leadership intent to execution, and where things started drifting apart.

How I work with organisations

There are usually three types of engagements.

1. Decision Drift Audit

This is a focused 10-day review.

I identify the institution’s most important decisions, map where drift risk exists, and show where intent and execution are no longer aligned.

At the end, leadership gets a practical, board-level summary of what’s happening, why it matters, and what needs attention first.

Most institutions discover the problem is bigger and older than they expected.

2. Decision Integrity Chain™ Forensic Review

This is deeper work, usually over 6 to 8 weeks.

It’s for situations where something has already gone wrong:

regulatory concerns
unexplained losses
governance failures
AI decisions nobody can fully reconstruct
controls that technically passed but still failed in practice

Here, I trace the full decision path across systems, teams, controls, models, and governance structures to identify where accountability and execution broke down.

3. Ongoing Advisory

Some institutions keep me involved longer term.

As systems become more automated and AI-driven, decisions happen faster and faster. Governance often struggles to keep up.

The goal here is to make sure decision integrity evolves alongside operational speed instead of falling behind it.

Patterns I keep seeing

After 25 years, certain patterns repeat themselves constantly.

Retail banking

Customers receive different decisions across digital, branch, and call-centre channels. Nobody planned the inconsistency. It simply built up over time through local fixes and process changes.

Private banking

Relationship managers leave during platform migrations, and the reasoning behind years of client decisions disappears with them. The data remains. The judgment history does not.

Treasury and deposits

Many funding models still assume customer balances behave the way they did ten years ago. But businesses now move money instantly using automated treasury tools and real-time yield platforms. The balance sheet may look stable while the actual behaviour underneath has completely changed.

AI-driven credit decisions

The model slowly shifts approval patterns over time. Every single change looks individually reasonable. But six months later, the overall lending behaviour no longer matches what leadership originally intended.

Operational risk

Manual overrides increase gradually until they stop being “exceptions” and quietly become normal business operations.

Healthcare operations

Processes become overloaded with validation layers and workarounds. Teams adapt locally just to keep work moving. Over time, revenue leakage and operational inefficiencies grow without any single obvious failure point.

What insitutions get from this work

At the end of an engagement, leadership gets a clear picture of:

where governance and execution separated
which decisions carry the highest drift risk
where accountability became unclear
how exposure built up over time
what actions need immediate attention

This is not a strategy presentation or a technology assessment.

It’s an honest examination of how decisions are actually being made inside the organisation versus how leadership believes they are being made.

And those are often two very different things.

Starting the conversation

The first step is usually a short discussion under NDA.

Most conversations begin with one uncomfortable question:

“Is the institution really making the decisions leadership thinks it’s making?”

That’s normally where the real work starts.

Decision Engineering™ — framework, autopsies, research → SSRN — six papers, eleven eJournals →