Every organisation hits a ceiling on how many good decisions it can make in a given period. Not because people aren’t smart. Not because data doesn’t exist. But because the work of turning data into decisions doesn’t scale the way other things do.

How decisions get made

Consider how a reasonably complex business decision happens today.

Someone notices a problem or opportunity. They request data. An analyst pulls a report, maybe several. The data arrives, but it doesn’t quite answer the question, so there’s back-and-forth. Eventually, enough context is assembled that a decision maker can weigh the trade-offs and make a call.

This process works. It’s how most organisations operate. But notice where the constraints are.

The analyst who knows how to pull the data is a bottleneck. There are only so many requests they can handle. The senior person who holds enough context to interpret the data is a bottleneck. Their calendar is full. The meeting where stakeholders align on what the numbers mean is a bottleneck. It takes a week to schedule.

None of these bottlenecks are about data availability. The data exists. The bottleneck is synthesis: the work of combining multiple sources, adding context, weighing trade-offs, and arriving at a recommendation.

Synthesis is skilled work. It requires judgment. It’s hard to delegate. And because it’s hard to delegate, it tends to concentrate in a small number of people who become organisational chokepoints.

The symptoms

You can recognise the synthesis bottleneck by its symptoms.

Decisions take longer than they should. Not because the decision is genuinely complex, but because assembling the evidence takes time. A question that should take an hour to answer takes two weeks because it has to queue behind other requests.

Gut feel fills the gap. When getting proper evidence is slow, people default to intuition. Sometimes intuition is right. Often it’s just faster. The organisation develops a culture of “we don’t have time to analyse this properly.”

The same questions get asked repeatedly. Different people, different contexts, same underlying question. Each time, someone rebuilds the analysis from scratch. There’s no leverage.

Heroic individuals become critical paths. The person who “knows where all the bodies are buried” becomes indispensable. This is good for that person’s job security, bad for organisational resilience. When they’re on holiday, decisions wait.

Senior time gets consumed by interpretation work. Leaders spend their days in meetings explaining what the numbers mean, when their time would be better spent on decisions that actually require their judgment.

If any of this sounds familiar, you’re experiencing the synthesis bottleneck.

Why dashboards don’t solve the problem

The last generation of business intelligence tools were built on a premise: if we give people visibility into their data, they’ll make better decisions.

This was true, as far as it went. Dashboards solved the visibility problem. You can now see, in near real-time, what’s happening across your business. That’s genuinely valuable.

But visibility is not the same as synthesis.

A dashboard shows you that a number is red. It doesn’t tell you why, or what to do about it. It certainly doesn’t weigh that red number against three other factors and recommend a course of action.

So what happens? People look at dashboards, notice problems, and then the synthesis process begins anyway. Pull additional context. Schedule a meeting. Build a spreadsheet that combines what the dashboard shows with what it doesn’t. The dashboard is a starting point, not an answer.

I’ve heard the same feedback from operators in different industries: “They build dashboards for us, and then we do nothing with them.” Not because the dashboards are bad. Because dashboards solve the wrong problem. When the constraint was “we can’t see what’s happening,” dashboards were the right tool. When the constraint is “we can see what’s happening but can’t process it all into decisions,” dashboards aren’t enough.

The cost

The synthesis bottleneck has real costs, even if they don’t show up on a balance sheet.

Speed. Competitors who can evaluate opportunities faster will capture them first. The time between “we should look into this” and “here’s what we’re doing” is a competitive variable.

Quality. Decisions made on gut feel or incomplete evidence are, on average, worse than decisions made with proper synthesis. The errors compound.

Talent. Your best people spend their time on synthesis work that doesn’t require their full capability. They’re overqualified for the task but necessary because no one else can do it.

Resilience. When synthesis capability concentrates in a few individuals, the organisation is fragile. People leave. People get sick. People go on holiday. The bottleneck becomes a single point of failure.

These costs are largely invisible because they’re opportunity costs. You don’t see the decisions that took too long, the opportunities that went to competitors, the talent that burned out on work beneath their capability.

The question

Organisations have more data than ever, better tools for storing and visualising it, and yet decisions still feel slow. The bottleneck moved, but it didn’t disappear.

The question is whether there’s a different approach. Not better dashboards. Not more analysts. Something that changes the scaling dynamics of synthesis itself.

I think there is. But it requires rethinking how we build data systems and what we expect AI to do for us.

More on that next.