You Can’t Fix What You Can’t See

Many African value chains aren’t broken because we lack ideas, ambition, or funding. They’re broken because no one has taken the time to trace the flow, spot the friction, and redesign the process for performance.

Over time, we’ve normalised systems that are held together by workarounds. A delivery delay becomes someone’s job to chase down. A paperwork bottleneck becomes a reason to hire another clerk. A mismatch between capacity and demand becomes just another day in operations.

These aren’t isolated issues. They’re symptoms of a deeper problem: we’ve stopped looking at systems as systems. We treat problems in isolation. We celebrate outputs, not flow. And we often reach for tools — dashboards, platforms, apps — as if they can fix what’s fundamentally a process problem. The problem isn’t the tools. The problem is when they’re thrown at a system without understanding it first. Without first asking what’s actually broken, and why, all we end up doing is digitising dysfunction.

That’s the missing step. The habit of stepping back, mapping how things really work, and asking:
Where is value being created — and where is it being lost?
What’s slowing us down, stretching us thin, or costing more than it should?
What are we doing out of habit, rather than by design?

This piece is about that perspective — about learning to look at life like an optimiser.

Not just in factories or value chains, but across teams, organisations, even public systems. Because once you know how to see friction, you start seeing opportunities. You start noticing misalignment. Waste. Delays that aren’t inevitable. Effort that isn’t adding value. And that’s when real change becomes possible.

In the first part of this series, we looked at the quiet logic of Operations Research — the original toolkit for decision-making, trade-offs, and resource allocation.

In the second, we broke down the AI toolbox — showing how different approaches can support prediction, optimisation, or content creation depending on the problem at hand.

This third piece brings us into the field.
Not into models or theories — but into what it means to observe, diagnose, and redesign a process in the real world.

Because productivity doesn’t start with tools.
It starts with how you look.

The Art of Seeing Flow (and Where It Breaks)

Before you optimise anything, you need to see how it actually works.

Not how it was designed to work. Not how it looks in a boardroom presentation. And definitely not how people describe it when asked. But how it really works — in motion, on a normal day, with all the delays, handoffs, and half-solutions that keep the system afloat.

This is what I mean by “seeing flow.”

It’s the habit of tracing how value moves through a system — how goods are handled, how information travels, how tasks connect, how decisions get made. And more importantly, it’s the ability to notice where that flow gets blocked, delayed, rerouted, or quietly drained.

Some people are naturally wired to spot these things — and I’ll be honest: I’m one of them.

As an optimiser, I can’t not see inefficiencies. I walk into a workspace and immediately notice where time is being wasted, where tasks don’t line up, where flow breaks down. And I’ve often wondered: why doesn’t everyone else see it?
This article is, in part, a quiet plea — a hope that once you learn to see what I see, the work of fixing it won’t feel quite so lonely.

Because once you start noticing flow — and where it breaks — you can’t unsee it.

And the friction is everywhere, if you start looking:

• Queues forming at one stage while another stands idle
• Information gaps, where people are waiting for signoff, specs, or stock updates
• Redundant steps — like filling out a form no one reads, or copying something between systems that don’t talk to each other
• Manual interventions that keep the process running — but shouldn’t be needed in the first place

These aren’t minor issues. They’re signals. They tell you where the system is leaking energy — or where it’s being held together by human effort and improvisation.

There’s a reason some of the best-run companies ask their executives to spend time in operations.
Put a CEO on the factory floor for a day, and two things usually happen.

First, they realise how many well-meaning decisions made in the boardroom are quietly creating friction downstream — adding steps, creating delays, or making life harder for the people trying to deliver.

Second, they discover issues that never even made it to leadership’s radar: small things, like a confusing form or a missing tool, that are quietly costing time and morale every single day.

Those details don’t show up on dashboards. But they shape how the system performs — and whether it improves over time or slowly grinds down.

That’s why I spend time in the system.
And why I hope you will too.
Because you can’t redesign what you’ve never taken the time to understand.

What a Factory Floor Can Tell You (Even Without Data)

People often think optimisation is about numbers. That to improve a system, you first need software, analytics, and perfect visibility.

But some of the clearest insights don’t come from data dashboards. They come from standing in the right place, watching carefully, and asking a few well-placed questions.

You can learn a lot from a factory floor. Or a warehouse. Or a back office. Even a school kitchen or a rural dispensary. Because every system leaves a trail — of decisions made, time spent, effort applied, and workarounds created.

Most of the time, nobody’s really watching the whole thing end to end. People are focused on their own task. Their own station. Their own targets. But flow doesn’t happen in isolation — it’s what connects all those tasks together.

If you’re willing to walk through slowly, without interrupting or judging, here’s what to pay attention to:

• Where is time being lost? Not in the dramatic failures — but in the small, repeated delays: looking for a tool, waiting for a signature, walking between departments, restarting a machine.
• Where is work piling up? If items or paperwork are sitting in stacks, it often means the next stage isn’t ready — or isn’t designed to absorb that volume smoothly.
• Where are people creating their own systems? You’ll find spreadsheets outside the ERP, hand-drawn forms next to digital screens, or clever hacks to “make it work.” These are gold — they tell you what the system should be doing but isn’t.
• Where does information stop flowing? A breakdown in communication is often a bigger constraint than a shortage of people or machines.
• What’s being done manually that could be avoided entirely? This isn’t about automation for the sake of it. It’s about tasks that shouldn’t need to exist if the upstream flow was better designed.

None of this requires a consultant’s report or a 12-month diagnostic.
It requires observation. Curiosity. And the humility to assume the system isn’t fine just because nobody’s complaining.
(And if you do want to outsource it — well, make sure you give it to an optimiser like me.)

Because in many organisations, people have adapted so well to the dysfunction, they no longer flag it. They just cope. They find workarounds. They make it work.
But the cost of those workarounds? Time. Energy. Money. Often invisibly.

So don’t just look for problems.
Look for effort that shouldn’t be needed. Look for energy that’s being spent holding the system together, rather than moving it forward.

That’s where optimisation begins.

Diagnosing Productivity Gaps — The Three Leaks

Once you start seeing flow, you’ll also start seeing where it leaks.

It’s rarely one big breakdown that’s slowing a system down. It’s usually small, consistent points of loss — moments where energy, time, or resources slip through the cracks. And the tricky part? Most of it doesn’t look like failure. It just looks… normal. Accepted. Built in.

Over time, I’ve found that most productivity losses can be traced back to one of three types of leaks:

1. Time Leaks

These are the moments that quietly drain hours from the day without anyone noticing.

You’ll spot them in the form of waiting: waiting for approval, waiting for deliveries, waiting for machines to warm up, or for colleagues to respond. You’ll also find them in switching costs — moving between tasks or locations that aren’t logically sequenced.

Ask: Where are people idle not because they’re slow, but because the system is?
Could a change in sequence, batching, or timing prevent the wait in the first place?

2. Capacity Leaks

This is when resources are available — but not used effectively.

A line might be overstaffed at one stage and under-resourced at another. A machine might be perfectly functional but underused because no one scheduled it properly. A truck might return half-empty because no one planned the reverse route.

Ask: Are we matching the right resources to the right need, at the right time?
And are constraints real — or just inherited from how things have always been done?

3. Information Leaks

The most invisible of the three — and often the most expensive.

Work slows down not because people don’t want to move, but because they don’t know what’s next. They don’t have the latest data. The spec didn’t get passed on. The stock level is out of sync. A supplier changed terms and no one updated the system.

Ask: Where does information stop flowing — or arrive too late to be useful?
And what decisions are being made with outdated, incomplete, or unclear data?

These three leaks — time, capacity, and information — are at the root of most performance issues I’ve seen.

And here’s the real trick: you don’t need perfect systems to fix them.
You just need to know how to name them. Because once you name them, you can start to test small shifts — and see what changes.

Sometimes that means moving a task upstream.
Sometimes it means removing a step altogether.
And sometimes, it just means giving people the right information earlier.

Optimisation isn’t about perfection. It’s about tightening the flow — so less gets lost.

Leverage Points — What to Fix First

Once you start noticing everything that’s not working, it can get overwhelming. Suddenly, the whole system looks broken. Everything feels urgent. But trying to fix everything at once is a guaranteed way to fix nothing at all.

Optimisation isn’t about doing more. It’s about doing less, but with more impact.

The key is to find your leverage points — those small changes that unlock outsized improvements. And that starts by knowing what not to fix first.

❌ Don’t start with what’s most visible.

The loudest problems are often just symptoms. A slow delivery isn’t always a logistics issue — it might be a stock visibility issue, or a batching delay two steps earlier.

❌ Don’t start with what’s easiest.

Quick wins are tempting, but if they’re in low-impact parts of the system, they won’t move the needle. Worse, they’ll drain your momentum.

👉🏽 Instead, look for:

1. The Bottleneck

Every system has a constraint — the stage that limits overall output. Improving anything upstream or downstream won’t matter if this point stays blocked.

Ask: Where is the real constraint?
If we could speed up just one step, where would it unlock flow across everything else?

2. The Energy Sink

Some parts of the system require enormous effort — without generating equivalent value. They’re exhausting to maintain. They drain teams. And they’re often held together by manual intervention.

Ask: What part of the system feels harder than it should?
Could we simplify, automate, or rethink it altogether?

3. The Repeated Workaround

If people are constantly bypassing the system, the system is the problem.

Ask: What do people keep doing “on the side” to make things work?
What would need to change so they didn’t have to?

Start there. With the bottleneck. The energy sink. The workaround.
Not because they’re easy — but because they unlock flow.
And flow is what makes everything else easier to improve.

When to Bring in the Tools (OR, AI, or Just Better Process Design)

Once you’ve mapped the flow and spotted the leaks, the next question is usually: what now?

This is where people tend to jump straight to solutions — bring in software, automate a process, build a dashboard, run a model. But the real question isn’t what tool do we use?
It’s: what problem are we actually trying to solve?

Because different problems call for different tools — and the wrong tool, applied well, can still waste your time.

🔸 Use Operations Research (OR)

When you need to decide between options under constraints.

OR is your go-to when you’re dealing with trade-offs: costs vs. time, capacity vs. demand, routes vs. fuel, inventory vs. service levels. It’s not here to “learn” — it’s here to calculate the best possible outcome based on the rules you give it.

Use it for:

·       Routing, scheduling, and resource allocation across logistics, production, or workforce planning

·       Balancing inventory, service levels, and operating costs in supply chains

·       Optimising operational processes and value chains for better flow and productivity

Ask: Where do we need a better decision, not just more data?
And do we understand the constraints clearly enough to model them?

🔸 Use Prediction-Based AI

When you need to anticipate what’s likely to happen.

This is where machine learning thrives — spotting patterns in data, forecasting outcomes, flagging risks. It’s helpful in environments with uncertainty and lots of small decisions.

Use it for:

·       Forecasting demand or spoilage

·       Predicting delays, equipment failure, or customer churn

·       Flagging anomalies or early signs of risk

Ask: What are we constantly reacting to — that we could start predicting instead?
And do we have the data needed to train that prediction?

🔸 Use Generative AI

When you need speed, scale, or variation in content creation.

It won’t optimise your supply chain — but it can write SOPs, generate training materials, or create ten versions of a customer message in minutes. Think of it as your prototyping assistant.

Use it for:

·       Drafting operations manuals

·       Creating internal training content

·       Simulating stakeholder scenarios or communication materials

Ask: Where are we slowing down because someone has to “sit and write”?
Could we start with a draft, then refine?

🔹 And Sometimes… Just Use Better Process Design

Not everything needs a model or a machine learning algorithm. Sometimes the best tool is a marker pen and a whiteboard. Sitting down with the people in the system, mapping the process, asking why each step exists — and removing what no longer serves a purpose.

Ask: If we had to redesign this from scratch, would we do it the same way?
And if not — what’s stopping us?

Tool choice doesn’t come first. Problem clarity does.

Once you’ve observed the system, spotted the friction, and understood what kind of gap you’re dealing with — then you reach into the toolbox. And if you’ve done the earlier steps well, you’ll often find that the right solution is smaller, simpler, and closer than you thought.

Because the best tools don’t just solve the problem.
They make the whole system feel… lighter.

(If you want to go deeper into how these different types of AI work — and how to choose between them — check out the second article in this series: “The Ultimate Guide to AI for Corporate.”)

From Seeing to Doing

The goal of this article wasn’t to give you a framework or a checklist. It was to help you shift how you look at systems — and maybe, how you move through them.

Because once you learn to look at life like an optimiser, the world stops feeling random. You start seeing why things stall. Why good people burn out. Why outcomes don’t match effort.
And more importantly — you start seeing where to intervene.

You don’t need to be a data scientist.
You don’t need to build a model.
You just need to slow down long enough to watch the system.
Not the reports. Not the results. But the system that produces them.

Start by asking simple questions:

• Where is time being lost?
• Where is effort being wasted?
• Where is flow breaking down?
• And what are we doing just because we’ve always done it that way?

The answers will rarely show up in bold.
But they will show up — in queues, delays, workarounds, and missed handoffs. In the tired faces of frontline staff. In the places where people are keeping things going, not because the system supports them… but in spite of it.

That’s the opportunity.

We don’t need more dashboards per se.
We need smarter dashboards — ones that reflect the right constraints, trade-offs, and priorities for your business. Not just the generic KPIs that everyone else is tracking.
And we need people who know what those dashboards should be showing — and why.

Because in the end, the tools don’t optimise anything on their own.

People do.

Not the ones who add layers of noise.
The ones who trace the friction. Read the system. Ask the uncomfortable questions. And design around what’s reallygoing on.

That’s the kind of entrepreneurship, leadership, and transformation we need.
Not just vision. Not just innovation. But optimisation that sticks ✨