EOS teaches leadership teams to run their businesses on data, not feelings. But does that mean we should stop trusting our judgement?
I recently had a fascinating conversation with Ric Segal of Founders Keepers about AI, data and decision-making.
As we discussed the extraordinary ability AI increasingly gives us to gather, analyse and interpret information, the conversation moved naturally towards a more human question:
What happens when the data and our judgement disagree?
Ric pointed me towards two shareholder letters from Jeff Bezos that explore exactly this tension.
And they got me thinking about EOS.
“The heavy lifting is done by the math”
Amazon is hardly an organisation known for ignoring data.
In his 2005 shareholder letter, Bezos described decisions such as where to locate fulfilment centres and how much inventory to hold. Amazon could model these decisions using historical information, costs, demand and other variables.
In those situations, Bezos was unequivocal:
“The heavy lifting is done by the math.”
Judgement still exists, but it is the junior partner.
That feels very familiar to anyone running EOS.
The Data Component asks us to boil the business down to a handful of objective numbers that give us an absolute pulse on the business.
The Scorecard helps remove ego, emotion and opinion from conversations about performance.
Rather than:
“I think sales activity is pretty good.”
We have:
“We generated 17 qualified leads against a weekly goal of 25.”
One is an opinion. The other is a fact.
And that discipline is enormously powerful.
But Bezos doesn’t stop there.
What happens when the maths can’t give you the answer?
In the same letter, Bezos describes another category of decision.
Sometimes there isn’t enough historical information. Sometimes the thing you’re deciding is genuinely new. Sometimes an experiment can’t tell you the answer before you have to make the decision.
In those circumstances, Bezos argues that data and analysis still have a role, but judgement becomes the prime ingredient.
His example is Amazon’s approach to pricing.
Amazon’s short-term analysis could tell it something quite clearly: reducing prices generally didn’t generate enough additional volume in the short term to compensate for the lost margin.
The maths effectively said: don’t cut the price.
But that wasn’t the whole question Bezos was trying to answer.
What Amazon couldn’t reliably calculate was the cumulative effect over five or ten years of repeatedly passing efficiencies back to customers through lower prices.
Would that increase trust?
Would it change customer behaviour?
Would it strengthen Amazon’s competitive position?
Would it contribute to the flywheel?
The available data could answer the short-term question.
It couldn’t answer the long-term one.
So Amazon didn’t reject the data.
It recognised the boundary of what the data could tell it.
I think that’s an important distinction.
A Measurable is not reality
Bezos develops the idea further in his 2016 shareholder letter when he warns organisations about allowing proxies to replace the thing they’re actually trying to understand.
A process can become a proxy for an outcome.
A customer survey can become a proxy for genuinely understanding the customer.
And, I would suggest, a metric can become a proxy for reality.
That’s where this becomes particularly interesting through an EOS lens.
An EOS Scorecard gives us an extraordinarily useful weekly pulse on a business.
But the Scorecard isn’t the business.
Your 5–15 Measurables are things the leadership team has deliberately chosen because it believes they provide a useful, predictive view of how the business is performing.
That’s precisely why they’re valuable.
But choosing what to measure necessarily means choosing what not to measure.
So what happens when every relevant number is green, but an experienced member of the leadership team says:
“Something doesn’t feel right.”
Don’t choose between the Scorecard and your gut
I think there are two tempting responses.
The first is:
“The numbers are green. There’s no problem.”
The second is:
“I trust my gut. The numbers must be wrong.”
Neither feels satisfactory.
And, importantly, I don’t think either is Pure EOS.
The number is the number. Don’t manipulate it, reinterpret it or dismiss it because somebody doesn’t like what it says.
But equally, don’t use a green number to prevent someone from raising an Issue.
Because an Issue in EOS isn’t simply a red number.
It can be a problem, an idea, an opportunity or a risk.
So perhaps the right response is:
“The Scorecard says we’re on track. Your judgement says something may be wrong. That’s an Issue. Let’s IDS it.”
That preserves the integrity of both.
Gut is a signal, not proof
This is where I think we need to be careful.
None of this means:
“Trust your gut.”
Gut instinct can be wrong.
It can be ego.
It can be fear.
It can be confirmation bias.
It can be a Visionary who has already decided what they want to do and would rather not have inconvenient numbers getting in the way.
Data is one of our best protections against exactly that kind of leadership.
But experienced judgement can also detect patterns before we have consciously articulated them.
A customer conversation feels different.
Three good people leave within six months.
The numbers still look healthy, but the energy in the business has changed.
A competitor starts doing something that initially appears insignificant.
A Visionary or Integrator notices something that isn’t yet represented anywhere on the Scorecard.
The feeling itself doesn’t establish that something is wrong.
But it may be enough to ask whether there is something worth identifying.
That’s a very different standard.
IDS provides the bridge
This may be where EOS already resolves the apparent tension between data and intuition.
We don’t need another tool.
We need to use the existing ones properly.
The Data Component gives us objectivity.
The Issues Component gives us somewhere to put the things that objectivity hasn’t yet explained.
And IDS gives us a disciplined way to interrogate them.
If the Scorecard is green but someone believes something is wrong, put it on the Issues List.
Then Identify.
What are we actually seeing?
What evidence supports it?
Are several apparently isolated anecdotes pointing towards the same thing?
Are we measuring the right number?
Is the goal right?
Is our Measurable still leading?
Has something changed in the environment?
Is the data itself accurate, complete and being reported consistently?
Or, having examined it properly, was the gut instinct simply wrong?
Any of those is a legitimate outcome.
The point isn’t that judgement defeats data.
The point is that disagreement between the two is itself information.
Sometimes judgement reveals something we haven’t chosen to measure.
Sometimes it tells us that a Measurable may no longer be the right one.
And sometimes it prompts us to double-check the data we’re receiving, because the problem may not be with the conclusion, but with the quality of the information underneath it.
AI makes this more important, not less
This is ultimately what brought Ric and me to the question in the first place.
AI is making extraordinary analytical capability available to increasingly ordinary businesses.
We can interrogate larger datasets, find patterns humans might miss, summarise customer feedback, model scenarios and make previously inaccessible analysis available in seconds.
That’s a huge opportunity.
But perhaps it makes one leadership capability more important rather than less:
knowing which questions the data can actually answer.
Better analysis of the wrong proxy doesn’t necessarily get us closer to the truth.
More data doesn’t eliminate the need for judgement about what matters.
And increasingly sophisticated AI doesn’t remove the human responsibility to be curious when something doesn’t make sense.
Perhaps the leadership skill isn’t choosing between data and gut at all.
It’s knowing how to use each appropriately.
Trust the Scorecard. But keep listening.
My takeaway is therefore not that EOS needs modifying.
Quite the opposite.
Pure EOS already gives us the disciplines we need.
Use the Scorecard to create an objective pulse on the business.
Trust the numbers rather than substituting opinions for facts.
But don’t confuse the Scorecard with the business itself.
If a customer comment, an observation, experience or even a nagging feeling suggests something important isn’t being captured, don’t override the data.
Raise the Issue.
Then IDS it.
Maybe the data will prove the intuition wrong.
Maybe the intuition will expose a flaw in what you’re measuring.
Maybe it will reveal that the data itself needs checking.
And occasionally, perhaps, it will identify something important before the numbers are capable of seeing it.
Data tells us what we’ve chosen to measure. Judgement helps us notice what we haven’t chosen to measure, or when we need to double-check the data we’re receiving.
The best leadership teams need both.
With thanks to Ric Segal of Founders Keepers for the conversation that prompted this thinking, and particularly for pointing me towards Jeff Bezos’s 2005 and 2016 shareholder letters.
Further reading
Jeff Bezos, 2005 Amazon shareholder letter, decisions driven by mathematics versus decisions requiring judgement.
Jeff Bezos, 2016 Amazon shareholder letter, resisting proxies, customer intuition and high-velocity decision-making.