Leadership Success: Are You Earning It, or Borrowing It?

The article argues that leaders should imagine stripping away all organisational advantages such as brand trust, distribution, and authentication, and ask if the product would still attract users on its own in a competitive market. Only then can genuine product driven success be separated from the parent company's halo effect.

1. Two million users

A product review lands on your desk. It shows 2 million users, and the team presenting is proud. Rightly so. Two million is a large number by any measure.

Then you look at the parent organisation. It has 8 million active customers, and the feature has been sitting prominently inside the banking app in front of every one of them.

The interesting number was never 2 million. It was 25 per cent. And even that does not tell you whether those customers chose the product, value it, or would go looking for it if it vanished tomorrow.

You are the leader in that room, and you have two jobs. The first is to work out what the team actually achieved. The second is harder, and it is the one this article ends on: to work out the same thing about yourself.

Did the product succeed because it is good, or because it was carried by an organisation that was already successful?

2. Two products, one slide

Imagine two teams presenting at the same review.

Product A sits inside a bank with 8 million active customers. A new feature is placed prominently in the existing app, and 2 million customers use it. The team reports, quite honestly, that 2 million people use their product.

Product B is a standalone startup. Nobody is forced to download anything. Nobody already has an account. It still acquires 200,000 customers who found it themselves, installed it, registered, used it, and came back.

On a slide, Product A looks ten times more successful. Product B may have demonstrated something far more valuable: independent demand.

Product A inherited distribution, trust, identity, authentication, payments, an installed customer base, marketing reach, and a privileged position inside an app people already open every day. Product B had to earn every customer from a standing start with none of that.

I should be fair about the comparison. Product B’s 200,000 has no denominator either, and if I am going to complain that 2 million hides a 25 per cent conversion rate, I have to admit that 200,000 hides a story about impressions, app store traffic, and paid acquisition that I have not told. The point is not that the startup number is cleaner. It is that neither number means anything until you know what it was drawn from.

So the real question becomes: how much of your success did your product create, and how much did your organisation lend you?

3. The organisational halo

There is an old idea in psychology that explains part of this.

In 1920 Edward Thorndike asked commanding officers to rate soldiers on four separate qualities: physique, intelligence, leadership, and character. The ratings across these supposedly independent traits were correlated far more tightly than reality could plausibly explain. An officer’s rating on one trait bled into every other. They were not judging each quality in isolation. They were colouring every score with one general impression of whether the man was, overall, good or poor. Thorndike called it the halo effect.

I am stretching his finding when I apply it to organisations, and it is worth saying so. Thorndike described a bias inside a single rater’s head. What happens to a feature inside a successful app is closer to brand extension: trust built somewhere else transferring to something new. Different mechanism, same shape. A successful organisation creates a halo around every product placed inside it.

A feature inside a large banking app, retail platform, or social network benefits from that halo before its team has done anything at all. The customer already trusts the brand, already has the login, already has the habit of opening the app. Some of that trust and habit transfers to whatever appears in front of them, whether or not the thing itself deserves it.

None of this criticises the product team. It is a warning about the measurement sitting on top of their work.

4. Inherited success, or earned success

The useful distinction is not succeeding because of an organisation versus succeeding despite one. A genuinely excellent product inside a well run organisation succeeds because of both. The fault line is success the organisation lent the product versus success the product created on its own.

Some products succeed largely on inherited advantage. They draw on distribution, an existing customer base, brand trust, prime placement, and payments infrastructure someone else built and paid for. Put almost anything useful enough in front of 8 million existing customers and somebody will use it.

Other products succeed on demand they created. Customers seek them out, tell others without being asked, come back without a push notification, and find workarounds when the organisation makes the experience clumsy. That demand survives poor placement, thin marketing, and outright neglect. None of it was lent.

On a dashboard these two look identical. When the organisation is exceptionally successful, they become almost impossible to tell apart from the inside.

So here is the thought experiment I would put in front of any team celebrating a big user number.

Take the product out of the mothership. Give it a new name and a new app. Strip away the existing login, the homepage placement, the push notifications, the customer database, and the trusted brand. Put what is left in an app store next to twenty competitors doing something similar. Would anybody download it? Harder still, would anybody pay for it?

This is a diagnostic, not a standard. Plenty of features are genuinely better for living inside an existing app, and integration itself creates real value for customers. The test cuts the other way too, and that direction is easier to get wrong. A leader who applies it too literally can starve an excellent, well integrated product of investment because it fails a standalone experiment it was never designed to pass. A fraud detection layer, an identity check woven into a verification pipeline, a feature that only makes sense alongside a customer’s full transaction history — each will look weak stripped of context. Not because it lacks merit, but because its merit is inseparable from where it sits. Defunding something like that mistakes an unfair question for a fair answer.

Often the honest answer is that almost nobody would use it independently. That is fine, as long as the team stops claiming 2 million users proves exceptional product market fit. What it proves is that the product and the organisation’s distribution together produced 2 million users. Those are different claims, and only one of them is about the product.

5. Three tests, and the number they correct

This is the part most product reviews skip, and the part that matters.

Start with plain exposure conversion: active users divided by meaningfully exposed users. In our example, 2 million over 8 million, or 25 per cent. Useful, but not yet earned adoption. If you put something in front of 8 million people and a quarter click it, that tells you about the strength of the placement as much as the strength of the product. Three checks turn that raw ratio into something closer to the truth.

Remove the promotion, and watch retention. Turn off the notification. Remove the homepage placement for a defined window. Watch daily and weekly return visits with no reminders in between. Set the threshold before you run it: what share of usage at 30, 90, and 180 days would you accept as evidence the product stands on its own? Decide that in advance, or you will rationalise whatever number arrives. A product people value keeps a meaningful share. A product that exists only because it keeps getting pushed collapses almost immediately. Retention that survives silence is the closest thing to proof.

Measure voluntary and organic discovery. Look at how customers actually arrived. How many came through a homepage banner, a forced upgrade screen, or a push notification, against how many searched for the product by name, followed a link from another customer, or typed it into an app store search box. A product with genuine pull shows a rising share of the second group over time. A product dependent on placement shows a flat or falling share, however the total is trending.

Ask the counterfactual. The hardest, and the one that matters most. Every serious causal measurement approach, from marketing incrementality testing to product holdouts, reduces to the same question: what would have happened if this had never existed? Apply it directly. Not “how many people used this,” but “how much of this behaviour would not exist without it.”

None of these require enormous investment. They require a willingness to measure the uncomfortable thing rather than the reportable one. Taken together — voluntary uptake, unprompted retention, organic return — they are what I mean by earned adoption. No single one of them is enough alone.

Exposure tells you how many people were shown the door. Earned adoption tells you how many walked through it and kept coming back.

And then decide something. A diagnosis nobody acts on is just a more sophisticated slide. If a product turns out to be largely inherited, that is rarely a reason to kill it — inherited distribution is a real asset, honestly deployed. It is a reason to change three things: reset the team’s targets so the next increment has to come from earned demand rather than more placement; stop citing the aggregate number as evidence of product quality, internally and externally; and be explicit that the product’s value is contingent on the platform, so nobody builds a strategy on the assumption it could stand alone. If a product turns out to be largely earned, fund it harder than the raw numbers justify. It has shown you something the dashboard cannot.

6. When the product and the ecosystem cannot be separated

All of the above assumes inherited and earned success are two things sitting side by side, waiting for the right test to prise them apart. In a mature ecosystem that assumption sometimes fails.

Some products would not exist in their current form without the ecosystem underneath them. A fraud model built on years of transaction data, an identity check built on an existing verification pipeline, a payments feature built on rails the bank already operates — none can be meaningfully imagined standalone, because the ecosystem is not distributing them. It is a structural ingredient of what they are.

The tests also carry some of the halo they are meant to filter out. A customer who searches for a feature by name still trusts the bank enough to think it worth searching for. Retention that survives the removal of a prompt still benefits from authentication that never fails and a support line that answers. And removing a promotion is itself a change the customer notices, so a drop in usage may reflect a disrupted habit rather than absent demand, while no drop may reflect a customer who has not yet noticed.

This is not an argument for abandoning the measurement. It is an argument for humility about what it proves. In a genuinely entangled case the honest conclusion is not a confident percentage split. It is that some of the success is held jointly, and that forcing a clean number onto it produces false precision rather than insight.

The risk sits on both sides of that admission. Claim too much precision and you punish a good product that happens to be structurally embedded. Claim too little and you let a mediocre one hide behind the argument that its success cannot be isolated. The tests narrow the uncertainty. They do not remove it, and knowing the difference is part of doing the measurement honestly.

7. The rust risk

There is a darker version of the same entanglement problem, and it matters more than the measurement question, because it explains how a successful company can decline without ever understanding why.

Bad products rarely announce themselves. They behave like rust, not like a fire alarm.

Somewhere inside a large product suite, a flow occasionally fails partway through, tells the customer something has gone wrong, and leaves them to call in and sort it out. The team looks at the numbers and sees a failure rate of 0.4 per cent that month. Small enough to note as a known issue and carry into the next sprint.

Nought point four per cent sounds trivial in isolation. It is not trivial repeated every month against a growing customer base, because rust accumulates. The call centre grows a little, and nobody can quite say why, because the calls are spread across a dozen root causes that never appear together on one dashboard. Sentiment shifts, not because any single failure was catastrophic, but because a small percentage of a large number is still a large number of people, and every one of them had a bad afternoon they will remember.

Here is the part that connects back to the halo. None of those customers blame the specific screen, the specific feature, or the team that shipped it. They experience one app and one brand, so the bad experience attaches to the company.

The organisation absorbed the credit for products it merely hosted. It now absorbs the blame for them too. The same entanglement that made success hard to attribute makes decline just as hard to attribute, at exactly the moment attribution matters most.

That is the real cost of never separating inherited from earned. A company confident in its aggregate numbers — healthy revenue, healthy usage, a brand that still tests well — can be losing trust steadily underneath, one small defect at a time, with no single metric sharp enough to catch it before the call centre, the churn, and the sentiment scores all arrive at the same conclusion together. By then it is not a product problem. It is a company wide one, and it took years of 0.4 per cent to get there.

8. Successful organisations need higher bars

This is not an argument against celebrating product teams. It is an argument for making success harder to fake.

A weak or young organisation gives its products brutal, honest feedback by default. Nobody knows the brand, nobody has the app installed, nobody is waiting for the feature. If a product grows there, something in the product is working.

A strong organisation can accidentally hide a mediocre product, because its distribution machine is powerful enough to make almost anything look successful. The stronger the organisation becomes, the more rigorous its measurement needs to be to compensate. Otherwise every team eventually arrives with the same slide: millions of customers use our product.

Well, yes. Millions of customers use your organisation.

So I would apply a higher bar to teams sitting on exceptional distribution, not a lower one. If you already have 8 million customers, a trusted brand, and millions of daily sessions, getting 300,000 people to try something once might be a disappointing result rather than an impressive one. A team elsewhere with none of those advantages, getting 40,000 people to actively seek out what they built, may have achieved something remarkable. The denominator changes the story, and it should change how the story gets rewarded.

The same logic does not stop at products. Leaders inherit organisational advantage exactly as products do. A leader can arrive into a business with extraordinary engineers, enormous distribution, an established brand, abundant capital, and years of accumulated infrastructure already in place. Revenue grows. Usage grows. Availability stays excellent. It becomes very easy to conclude this happened because of strong leadership.

The identical counterfactual applies. What happened because of the leader, and what happened because they inherited a machine that was already running well before they arrived?

The shoulders of giants are wonderful places to build from. They hand you distribution, trust, a customer base, and infrastructure a startup would spend years assembling. There is nothing wrong with using every advantage an organisation gives you. But standing on the shoulders of a giant is not the same as being one, and the two get confused constantly inside organisations that have already succeeded at something else.

So the question is not how many people use the product. It is how many chose it, returned without being asked, valued it enough to tell someone else, and would genuinely miss it if it disappeared tomorrow.

And then the version of the question that is considerably less comfortable to ask. If you removed the organisation’s brand, its people, its distribution, its capital, and its momentum from underneath you, how much of the success you currently claim would still belong to you?