Hub-and-spoke data teams: the model that sounds right and works badly

August 27, 2026 · 5 min · Free, with full transcript

Hub-and-spoke has become the default answer when CDOs are asked how to balance central governance with business-unit agility. The reality in most organisations is slower decisions, diluted accountability, and data professionals caught between two bosses with conflicting priorities.

Listen to the podcast

4 min

Read the full transcript

Welcome to the Leaders Insights Podcast.

Today's episode, Hub and Spoke Data Teams, the model that sounds right and works badly.

Every CDO I talk to says the same three words when I ask about their operating model.

Hub and Spoke, central team in the middle, embedded analysts in the business units.

It sounds like the grown up answer.

Why are you telling me it quietly fails?

This sounds like the grown up answer is exactly the trap.

The article that kicked this off cites a 2026 survey where Hub and Spoke was the single most

named data operating model.

And in the same breath, the majority of those same organizations reported slower decisions

than before they adopted it.

That's not a rollout problem.

That's the model doing what it's designed to do.

Slower?

The whole pitch is you get central standards and local speed.

How do you end up slower than a plain, centralized team?

Two bosses, the embedded analyst, the person actually building the dashboard, reports functionally

to a business unit that wants an answer by Thursday and dotted line to a central data

office that wants governance, lineage, the audit trail.

Lineage just means being able to trace a number back to where it came from.

So every meaningful request now needs a negotiation between two people whose bonuses point in

opposite directions.

You didn't remove the bottleneck.

You gave it a second head.

But surely that tension is healthy.

Governance should slow down the cowboy who wants a number by Thursday.

Healthy tension needs someone who can end the argument.

In Hub and Spoke, nobody can.

The business unit head says, this is my analyst.

The CDO says, this is my standard.

Neither owns the outcome.

So when the number's wrong, watch what happens.

The business blames Central for being slow.

Central blames the business for going rogue.

And the analyst in the middle updates their CV.

That's the diluted accountability, the piece names.

I've seen it at a European insurer where the same customer churn figure, churn being

the share of customers who walk, came out three different ways in one board meeting.

Three teams, three definitions, all technically governed.

Name names or it didn't happen.

Who's actually admitted this publicly?

The clean public example is Airbnb, and it predates the current hype.

Years back, they killed a sprawling metrics mess by building what they called Minerva,

one place where a metric like active bookings is defined once and served everywhere.

The lesson people took was build a tool.

The real lesson was organizational.

They stopped letting every spoke define its own truth.

Contrast that with the average hub and spoke shop in 2026, where each business

unit still owns its own definitions and Central just coordinates.

Coordination is not authority.

So what do you do instead?

You can't re-centralize everything, the business units revolve and rightly.

You don't re-centralize people.

You centralized decisions and decentralized hands.

Split the two things hub and spoke welds together.

Definitions, metrics, data contracts.

A data contract is just a written promise about what a data set contains and won't break.

Those live centrally, non-negotiable, one owner.

The building, the analysis, the answer by Thursday that lives fully in the business unit,

single boss, no dotted line.

The analyst reports to the business, full stop.

Then what stops your cowboy from inventing his own churn number again?

He can build whatever he wants, but he has to build it on the certified definition.

He can't redefine churn.

He can only use it.

That's the swap.

Freedom on the how.

Zero freedom on the what counts.

Vendors selling data catalog software, Atlan Calibra, will tell you their tool fixes this.

It doesn't.

A catalog is a filing cabinet.

It'll happily store your three conflicting churn definitions in tidy alphabetical order.

The fix is deciding who's allowed to define the word.

And that's a management call, not a purchase.

Give me the one thing a CDO listening right now does Monday morning.

Pick your five most argued-about metrics.

Revenue, churn, active user, whatever your board fights over.

For each one, write a single name next to it.

The person who owns the definition, not the dashboard.

If two names show up, or none, you've found your slowdown.

That's your Monday, not a reorg.

A list of five words and who owns them.

Five words and an owner.

That's cheaper than any reorg.

And it survives the reorg you'll do anyway.

That's it.

The CDO decision tools are live at MBA-training.com.

Read and go deeper