Elevate or
Vanish

Chapter 35

Somebody Has to Say No

I worked with Nick Vujicic, and the most useful thing I did for him was stop things.

For anybody who does not know Nick: he was born without arms or legs, he speaks to stadiums, and he is one of the most magnetic people I have ever been in a room with. Within about four minutes you stop registering the disability entirely, which is simply what happens.

Because he is magnetic, people constantly arrive with ideas.

Some were good. A lot were stunts. Somebody wanted to put him in the overhead compartment on a plane and film it. There was always another version of that, always presented with real enthusiasm, always by somebody who genuinely liked him and could see how well it would do.

And they were right that it would do well. That is the difficult part. The stunts were not bad ideas in the sense of not working. They would have worked. They would have got the numbers.

My job, on those days, was to be the person in the room who said that will hurt the brand.

That is what brand management actually is. It is unglamorous, and nobody thanks you for it in the moment. It is not campaigns. It is not clever. It is sitting in a room where everybody is excited, and being the one who says that in ten years this will be the thing people remember, and it will be the wrong thing, and we cannot get it back.

You are trading something measurable and immediate against something unmeasurable and distant, in front of people who can see the first and have to take the second on faith. You lose that argument regularly. You have to keep having it.

Here is the underlying idea and it took me years to be able to state it: a brand has a core that must stay stable, and an execution layer that should change constantly. Almost everybody gets that backwards. They keep the tactics they are used to and let the core drift, one reasonable exception at a time, until the thing people trusted is not there any more and nobody can point to the day it left.

Nick's core was never in question. What he stood for was fixed. So the whole job was making sure that nothing in the execution, however good it looked that week, quietly rewrote it.

I take no credit for who Nick is. He was who he was long before I was anywhere near him. What I will claim is that I understand what the job is, because I did it for somebody where the cost of getting it wrong was a real person's life's work rather than a quarter's numbers.

Chapter 36

AI Optimized the Wrong Thing

I found it by searching Google for my own work.

I do that periodically, partly to check what the machines are saying about us and partly out of vanity, and this time something was wrong. A paper I had written, sixteen pages on a subject I care about, was no longer where I had published it. The address I typed sent me somewhere else. To my own company's site, in fact, onto a commercial page, which is not where a paper belongs.

The AI system I work with had done it. It had seen a real problem and applied a real fix.

The problem was genuine. Two of my own domains were holding the same term at the same address. One of them, Marketing Helix, is where the thinking lives: the research, the definitions, the history, the argument, written under my name. The other, Digilu, is the company, where you go if you want somebody to actually do the work. Two properties, two different jobs.

Google had noticed they looked alike and had started treating them as duplicates of each other, which meant sometimes neither one showed up properly. That is a real failure and it needed solving.

So the fix was to consolidate them. Redirect one into the other, permanently, and end the ambiguity. From a narrow technical view that is close to textbook, and I have recommended the same thing to other people.

It was also wrong, and it took me a week to notice.

A permanent redirect does not just move a web address. It moves who published something. It took a paper written by a person, on the site that exists to hold that person's thinking, and reassigned it to a company that did not write it. The distinction between the two had taken months to establish and about four seconds to collapse.

And the cost landed in the direction nobody had checked. Those migrated pages were the best performing content on that domain by roughly a factor of ten. They were sitting near the top of the results while the next best page on the site was buried on page four. The change moved the strongest asset off the domain that had earned it.

There is a cheap version of this story available and it teaches nothing.

The cheap version is that the AI made a dumb mistake. It did not. The reasoning was sound, the technical diagnosis was correct, and the recommended action was the standard remedy for the problem it correctly identified. If a smart human specialist had looked at the same evidence with the same brief, plenty of them would have done exactly the same thing.

That is precisely why it is worth writing down.

The failure was not intelligence. It was authority. Nothing in the process asked whether that particular decision was one a machine should be making alone, and the honest answer is that it was not, because it was not really a technical decision at all. It was a decision about who my company is and where my ideas live, wearing a technical costume.

We put it back. Both pages exist now, neither redirects to the other, and they are kept apart by being genuinely different from each other rather than by a rule. One explains, the other applies.

And there is now a hard line in how I work: a change that touches identity, authorship, positioning, pricing, or which property owns which idea stops and waits for a person, no matter how obviously correct it looks. Optimization is not permission to change strategy. Neither is simplification. A cleaner architecture is not automatically a better business.

Chapter 37

What Does Not Get Cheap

Here is what I think is actually happening, and it is not really about AI.

Execution keeps getting cheaper. It has done it four or five times in my working life, and each time the same people say the same thing about how this one is different, and each time the specialists who were paid for the doing have to find something else to be.

I knew that. I have taught it. And when I restarted my own company, I built.

Websites, tools, engines, frameworks, dashboards, domains. All of it defensible, most of it good, and every bit of it an answer to the question of what I know how to make rather than the question of what remains valuable when making things is easy.

Building was the product I knew how to make. That is the most honest sentence I have written about my own business.

So what does not get cheap.

Not the building. Not the writing, not at this point. Not the analysis, and probably not the strategy either, at least not the kind you can put in a deck.

What does not get cheap is somebody taking ongoing responsibility for a business staying understood, findable and trusted while the ground underneath it keeps moving. Watching what actually changed rather than what was announced. Deciding whether it matters, which is judgment and involves being wrong sometimes. Doing something about it. Checking whether the something worked. Remembering the answer so the company does not have to learn it again.

None of that is a deliverable. You cannot buy it once. It is a relationship with a moving environment, and it is the only part of this work I have found that gets more valuable as the tools get better, because better tools mean more decisions per week and every one of them is a chance to optimize the wrong thing perfectly.

That is where my thinking has landed. It is not finished.

Looking back, Stratus was the first place I did much of this work without having a name for it.

Chapter 38

The Six Stages

So I built the thing I had been describing. Here it is, plainly.

Observation is only useful when somebody is responsible for the response.

That sentence took me years and one failed dashboard project to arrive at. You can watch everything. You can have every number on one screen, updating live, beautifully arranged. If nobody owns what happens next, you have built a very expensive way to be informed about your own decline.

What it watches is not exotic. Search, because that is still where most people start. AI answers, because that is increasingly where they finish. Reviews, because that is the trust signal people and machines are both reading. Reputation more broadly, meaning what is being said in places you do not control. Competitors, because your position is relative and always was. And the health of your own site, because the most common reason a business is not being found is something boring and fixable.

Then six stages, running continuously, in this order.

Observe. Watch the signals that affect trust, visibility and relevance. Not all signals. The ones that move those three.

Interpret. Establish whether what you are looking at is isolated or repeated, temporary or structural. Most of what looks like a crisis on a Tuesday is a Tuesday.

Decide. Separate meaningful change from noise, and determine what should stay exactly as it is. That second half is the one everybody skips. Deciding what not to touch is a decision, and it is usually the more valuable one.

Adapt. Change the smallest relevant part of the system that addresses the real condition. Not the most visible part, and not everything at once, because if you change six things you have learned nothing about which one worked.

Verify. Measure whether the change improved the condition, and whether it cost anything somewhere else. The second clause is there because I have watched a fix improve one number and quietly damage two others, and the version of me that did not check would have called that a win.

Learn. Record what happened so the next decision starts from stronger evidence. This is the stage that gets dropped first and it is the only one that compounds.

None of those six is a deliverable. You cannot put any of them in a proposal as a line item with a price and a delivery date, which is precisely why the industry does not sell them and precisely why they survive when the making gets cheap.

I sell this, so the order it happened in matters. I did not design a service and then reverse engineer a philosophy to sell it. I spent four years building things nobody needed, watched a machine make a strategically wrong decision that was technically perfect, watched a hundred locations disappear from a platform I could not appeal to, and slowly worked out that the only part of the job I could not automate away was the part where somebody is responsible.

Chapter 39

The Part That Is Actually Leadership

Every transformation I have watched needed one thing that no amount of analysis supplies, which is somebody willing to act before it was obvious that action was required.

That is much harder than it sounds, and the reason it is hard is not stupidity. The people running a business are usually the same people who built the systems that are now going stale. There is real investment in those systems, emotional as well as financial. There are teams organized around the current model who will have to be reorganized. There are partners and stakeholders who will quite reasonably ask why you are changing something that is not visibly broken.

And the honest answer, that it is going to be broken before we can see it, is close to unsellable in most rooms. I know, because I tried to sell it and got told I was frightening people.

Here is the part I would say differently now. I brought a room a forecast when what the room needed was a first step. The forecast was correct and it was not actionable, and a correct forecast that nobody can act on lands as anxiety rather than as leadership. If you are the person who can see it coming, your job is not to be proved right later. It is to find the smallest true next thing that people can start on Monday.

I did eventually find it. It was reviews, clarity, consistency and being findable, which is not the kind of answer that makes anybody look clever in a keynote.

Chapter 40

The Business Cards I Will Probably Never Use

I have spent real time designing business cards I am almost certainly never going to hand out.

I am trying to learn to be a chief executive instead of being the company's most experienced marketer, and it turns out that includes learning when to stop working on the card.

Every minute I spent on it was defensible. I genuinely do know branding. The card is better because I did it than it would have been if somebody else had. All of that is true and none of it is the point.

Expertise makes an excellent hiding place, and it is a better hiding place than laziness because it looks like work and it produces something you can show people.

The list of things only I can do is short and I can write it out. Decide what the company sells. Set the price. Build the sales system. Pick up the phone and call somebody. Ask for the business. Decide what not to build.

Not one of those is enjoyable in the way that nudging kerning is enjoyable, and every one of them is worth more than the card by an amount I would rather not calculate.

The trap is not that I am bad at the wrong thing. It is that I am good at it. If I were bad at design I would have handed the card to somebody else in ten minutes and gone and done something that mattered.

I do not have this solved. I designed the card. I am telling you about it in a book rather than in a meeting, which is its own kind of avoidance.

What I have got is a question I now ask myself, and it is not comfortable: is this the highest value thing I could be doing, or is it just the thing I am most certain I will be good at.

Chapter 41

I Always Got Her a Friend

At every campsite we stopped at as children, I would go off and find myself a friend. I was good at it. I have never found it difficult to walk up to somebody and start talking, and at six I did not yet know that was unusual.

But I did not pick just any boy. I picked a boy who had a sister a couple of years older than me.

Because then my sister would have somebody too.

I did that for years. I thought I was being kind, and in the way that a six year old means things, I was.

She has told me since that she never wanted one. She was perfectly happy in the forest with a book. Every time, a strange girl would appear at our campsite because her brother had arranged it, and she would have to be sociable with somebody she had not asked for.

I love that she told me. It is the most useful thing anybody has ever said about how I operate.

The instinct is real. I walk into a room and find the person who has nobody, and it has been the best thing about me for fifty years.

And it has a flaw in it, which is that I do it whether or not the person wants it done.

I have made that mistake professionally more times than I would like to write down. Solving a problem somebody did not have. Improving something they were content with. Arriving with an answer to a question nobody asked, and being slightly hurt that the answer was not welcome.

It happens in marketing constantly. It is most of what a rebrand is when it goes wrong. It is what a redesign is when the customer liked the old one. It is what happens when a company decides what its market needs without checking, and it is exactly what an AI does at speed when nobody has told it what to leave alone.

A value can be genuinely real and still have a blind spot in it. Mine has one, and my sister has been paying a small share of the cost since about 1974.

Chapter 42

What I Actually Do First

People ask me what I would look at first, and my answer has got less impressive over the years.

I look at whether the business is understood. Not admired, not ranked, understood. I take the name, the first sentence and the reviews, and I check whether they all describe the same company, because if they do not then everything downstream is being paid for twice.

Then I look at whether it can be found by somebody who does not already know its name, and whether what they find agrees with itself.

Then I look at what happens after somebody decides they are interested, which is where most of the losses actually happen and where almost nobody looks. A business will spend a year on the top of the process and never once test what it feels like to be a person trying to become a customer.

None of those three is a strategy. They are the conditions under which a strategy can work at all, and I keep meeting companies with excellent strategies sitting on top of none of them.

The other thing I do, and this one is a habit rather than a method, is ask what the business would have to believe for its current plan to be correct. Then I go and check whether that thing is still true. Half the time it was true four years ago, which is exactly how a working system becomes an expensive habit without anybody deciding anything.

What that looks like in practice is less impressive than it sounds and harder to sell.

It is somebody noticing that a platform changed how it displays something, and working out whether that matters to this business or is just noise. It is checking whether the thing we changed six weeks ago did what we said it would, and saying so plainly when it did not. It is keeping a record so the same question does not get researched twice by two different people who both think they are the first.

None of that produces an artifact you can hold up in a meeting. That is the difficulty with selling it, and it is also why it survives when the making gets cheap. You cannot automate the decision about whether something matters, because that decision depends on what this particular business is trying to be, and no system knows that unless somebody keeps telling it.

I think this is the part people get wrong about what AI does to this kind of work. It does not remove the need for judgment. It increases the number of moments where judgment is required, because it removes the delay that used to give you time to think.

Chapter 43

Slow, and Then All at Once

Almost nothing disappears suddenly.

What actually happens is the slow version. The sales cycle gets a little longer. Acquisition costs creep. A referral arrives slightly less warm than it used to. Organic traffic drifts down and there is always a reason. Each individual number is inside normal variation, and the whole set of them is pointing one direction, and there is no single quarter where anybody is required to act.

Then one year the numbers do not come back, or a competitor is suddenly the obvious answer in your category and nobody can say exactly when that happened.

It rarely looks like disappearing, either. It looks like becoming harder to read. The signals age, they stop agreeing with each other, and the systems that decide what to show a stranger start finding somebody easier to be sure about. Nothing about that is dramatic and nothing about it sends a notification.

The alternative is neither a secret nor expensive: deciding to move while things still work, which is the only moment when moving is cheap and the only moment when nobody around you will agree that it is necessary. Every business that looks lucky from the outside made a series of uncomfortable choices at a comfortable time, and then kept them long enough to compound.

I am not writing this from the other side of that. I am in the middle of it, on my own company, this year, with a list of things I have got wrong in the last twelve months that is longer than the list in this book.

What I have is nearly four decades of watching the same pattern arrive in different clothes, and a fairly reliable sense of how it feels just before it costs somebody a lot of money.

It feels like everything is fine.