1You will recognize this company
Growth has slowed. The board asks harder questions than it used to. Sales says the product story isn’t landing. Marketing says the positioning is stale. Somebody says the word “rebrand,” and because it is the only diagnosis on the table with a known treatment and a known price, it wins. An agency is hired. Workshops are run. A narrative emerges, then a visual identity, then a launch.
Six months later, nothing has changed. Win rates are flat. The sales team has quietly gone back to the old deck. The new language is on the website and nowhere else. The company has spent somewhere between $400,000 and $4 million to discover that it did not have a brand problem.
The cost of getting the diagnosis wrong extends well beyond marketing.
In Alvarez & Marsal’s 2026 survey of North American private equity investors, operating partners and portfolio executives, 41% reported realizing less than three-quarters of planned value creation over the prior twelve months. A&M describes a “diligence-to-execution gap.”
There are many reasons a value-creation plan can fall short in execution. This paper is concerned with one that is easy to miss. Sometimes the problem isn’t that the company failed to execute the plan. It’s that different parts of the company are executing different futures.
Product is building toward one future. Sales is selling another. The value-creation plan assumes a third. Marketing is then asked to make the whole thing sound coherent. That is not primarily a brand problem. It is a decision problem.
This paper argues that many problems diagnosed as brand failure, messaging failure or execution failure begin further upstream. Nobody has decided what business the company is becoming.
To see why nobody decides, start with the most respected idea in technology, the one nobody dares argue with.
2The customer obsession trap
Every technology company says the same thing. We put customers first. We work backwards from the customer. We are customer-obsessed.
There is a video that circulates every year or so, a supercut of enterprise technology adverts spliced together. Empowering your teams. Accelerating your journey. Your success is our mission. The joke is that you cannot tell where one company ends and the next begins. Everyone has seen it. Everyone laughs. Nobody asks the obvious question.
How did an industry this obsessed with customers end up this indistinguishable to them?
It is not because the companies are lying. It is because they are all telling the truth. The same truth. Ask customers what they want and they will tell you, honestly and in aggregate, the average demand of the category. Faster. Cheaper. More integrated. More secure. Build from those answers and you will produce, with precision, a description of your category. So will everyone else. The obsession is real. The sameness is its direct result.
Here is the sequence most technology companies actually run.
Listen to customers. Build what they ask for. Derive the story from what was built. Let what the company is emerge, by accumulation, from what it has shipped and promised. It feels virtuous. It feels humble. And it outsources the one decision that cannot be outsourced.
Consider what customers can actually give you. They can tell you what they need. They can rank your roadmap. They can tell you where the product disappoints, what they would pay, who they compared you against. This is real information, and companies that gather it well deserve their reputations.
But there is one question customers cannot answer, and that is what business you should become. Each customer can only answer as themselves, while the becoming question is about which customers you will choose, which you will decline, and which don’t exist yet. Put it to your customer base and the aggregate answer is always the same. Become more of what we already buy. Which is to say, stay put, in the middle of the category, with everyone else.
Here is the strange part. The becoming question is also the only question that makes customer answers useful. A company that has decided what it is becoming hears demand as signal. This request fits who we are. That one doesn’t. A company that hasn’t decided hears every request as strategy, and drifts, one reasonable accommodation at a time, into being a little of everything.
Customer obsession tells companies how to learn. It doesn’t tell them what to become.
What obsession cannot supply, someone must decide. Call it Commercial Identity, the commercial answer to one question. What company are we becoming? What we will become, what we will not, and how that choice builds on where this company can create disproportionate value. A company’s commercial choices are, in the end, what it actually believes. Everything else is what it says.
This is not product strategy, and the difference matters. Product strategy decides how products win. Commercial identity decides what company those products build. A strong product leader can answer the first brilliantly while the second stays unresolved around them.
The becoming question is not answered in a vacuum. It stands on a prior question. Where does disproportionate value actually come from in this business?
Every company has an engine, a place where advantage is actually created. For some it is the product. For others it is a scarce asset base, a distribution position, a network, proprietary knowledge, or the allocation of capital. Industry does not decide this. TSMC and NVIDIA sell into the same industry. TSMC’s value comes from fabs, process mastery and manufacturing capacity, an asset engine. NVIDIA owns no fabs at all. Its value comes from architecture and a software ecosystem, an IP engine. Same industry, different engines, and any identity worth having must be built on the right one.
This is the failure that never looks like one. A leadership team that misreads its engine can make a choice, align the company, and execute with discipline, and still be wrong, because the identity was built around someone else’s source of advantage. The culture of technology assumes every company is a product company. Some are. Many are not. An identity is only real if it is true to the source of the company’s advantage.
You can test whether a company has one. A real identity forces decisions. It tells product what not to build, sales who not to pursue, marketing which language is off-limits, corporate development which deals make no sense. If a statement of identity has no casualties, nothing the company will stop doing because of it, then it is not an identity. It is decoration.
And the sequence inverts.
Notice what moved. In the first sequence, the company comes last. What the business is arrives as a residue of everything it happened to ship and promise. In the second, that same question moves to the front and gets answered on purpose, and once it is a decision rather than a residue, it has a name. Choices first. Customers as the test of the choice, not the source of it. Language last.
3Messaging is the consequence
Now the brand problem this paper opened with explains itself.
When the sameness becomes visible, when win rates slip and the field says the story isn’t landing, companies reach for the visible layer. Messaging. New positioning, new narrative, new launch. But messaging is downstream. It can only express decisions that exist. Asked to compensate for decisions that don’t, it manufactures differentiation out of adjectives. You can hear it happen.
Run two tests this week.
The logo-swap test. Take your homepage headline and your top three product claims. Replace your logo with your closest competitor’s. If nothing reads false, you do not have a point of view. You have a description of your category.
The competitor-honesty test. For each claim, ask whether your closest competitor could honestly say the same words. Not whether they would. Whether they could, truthfully. A yes means the claim describes the category, not the company. The tell is benefit inflation. More, faster, smarter, scalable, trusted. When the words inflate, the decision upstream was never made.
Most technology companies fail both tests, and the failure gets blamed on marketing. It is not a marketing failure. Messaging teams are handed an unmade decision and asked to fix it with language. A new narrative every eighteen months, each wearing out on schedule, because language is being asked to do a job only a decision can do.
There is a sentence for the next meeting where a messaging project is proposed. “We are trying to solve this with messaging before we’ve decided what company we’re becoming.” If the room goes quiet, you have your diagnosis.
Weak commercial identity forces marketing to compete with language. Strong commercial identity allows marketing to compete with truth.
4AI doesn’t resolve incoherence. It industrializes it.
For decades, the cost of an unmade identity was survivable, because organizations are full of people who absorb it. Sellers sense the contradiction in the deck and pick a story. Solution architects read the room. Marketers write language designed to survive several futures at once. An enterprise is, among other things, a machine for metabolizing its own incoherence, through thousands of small human acts of interpretation, every day, at the edge.
That is the machine every company is now automating away.
Look at what is being deployed right now, and ask the questions as you go. A copilot is being trained on your sales content as it stands today. Which of the three incompatible descriptions of your ideal customer, accumulated across three strategy cycles, will it learn? An AI SDR starts prospecting tonight. Which era’s definition of your ideal customer is it working from? An assistant is fielding the question “what should I build on?” Which version of your company’s self-description, of all the ones ever uploaded to the knowledge base, will it choose?
A human who senses a contradiction picks a side and commits. A system trained on both futures does not choose between them. It blends them, deterministically, at scale, in front of customers.
AI doesn’t resolve commercial incoherence. It industrializes it.
No company is exempt. The large ones are just better at paying. When generative AI arrived, the cloud industry faced the question in its purest form. Are we a model company, or the home of every model? Each is a real identity with real casualties. One major cloud chose both. Whether that hedge will prove right is a question for cloud analysts, and this paper does not need the answer. What matters is what holding it costs. For two years, thousands of people have made it work, deal by deal, keynote by keynote, with language engineered to be true in both futures. An unmade decision at hyperscale is funded by an army of human interpreters and margins deep enough to pay them. The companies in most portfolios have neither. And they are automating away the few interpreters they have.
The window is real, but it is not a countdown. In the same A&M survey, 73% of private equity respondents expect AI to boost portfolio value within the next year, yet only 8% describe their own firm as leading in AI implementation. The encoding is ahead of most companies, not behind them. In BCG’s May 2026 survey, roughly 60% of CEOs said their boards are rushing AI transformation. The urgency is real, and it is anxiety-shaped. What remains is an ordering constraint. Coherence is cheap before you encode and expensive after, for the same reason data cleanup is cheap before the migration and brutal after it. The scarce resource in the boardroom is not AI enthusiasm. It is a decision about what the AI is executing for.
None of this depends on when, or whether, AGI arrives. Should the frontier stall tomorrow, companies will still spend the next decade wiring commercial logic into systems. The argument is about organizations, not models.
5Why the usual process can’t make the decision
If the diagnosis is an unmade decision, why doesn’t the standard process make it? The offsite, the brand workshop, the positioning sprint.
Because the standard process is built backwards. It opens with a workshop, and workshops are alignment mechanisms, not discovery mechanisms. Put twelve executives in a room and ask what the company should become, and you will get the intersection of what everyone can accept, which is, by construction, the category average. Consensus is the wrong instrument for a decision whose value lies in its casualties.
The information you need is not in the room’s agreement. It is in its disagreement. Interview the same twelve executives one at a time and a different company appears in each conversation. The CRO is selling one business. Product is building a second. The CFO is underwriting a third. The board deck describes a fourth. Each version is being executed in good faith, right now.
The purpose of executive interviews isn’t to gather opinions. It’s to expose contradictions. The disagreement is the diagnostic.
So the working sequence inverts the standard one. Interviews first, one to one. Contradictions mapped and attributed to choices, not to people. Then leadership faces the actual decision in the form the disagreement revealed. Not “what’s our story?” but “three businesses are being built under one logo. Which one are we?” Only after that choice is made does anyone touch messaging. The workshop, if there is one, comes last, where alignment belongs.
And the decision alone is not the finish line. A distinctive choice the organization never propagates is a stranded strategy. Coherent execution of an undistinctive choice is an efficient commodity. Advantage compounds only when both are present, which is why the work runs from the choice all the way through to the systems that will execute it.
One honest boundary. Not every company has an identity problem. Some have a pricing problem, a product problem, or a distribution problem, and a real diagnostic will say so and stop. A process that cannot conclude “you don’t need this” is not a diagnostic. It is a sales pitch.
6The decision
If you run a company, the identity question cannot be delegated. Not to marketing, not to an agency, not to a workshop, and, from now on, not to the systems that will otherwise answer it by default. Suppose your executive team wrote down, separately, what business the company is becoming. How many businesses would you be running?
If you sit on a board, the governance question is not how fast the company is moving on AI. It is what the company has decided the AI will execute. Slowing down to answer that is not caution. It is sequencing.
If you invest, walk the portfolio with this lens and count the companies building two or three businesses under one logo, with a value-creation plan underwritten against a fourth. They are about to spend your capital encoding the ambiguity. And the diligence question worth adding is simple. Where does disproportionate value originate in this company, and is the identity built on it?
Every technology company is about to encode how it competes.
AI won’t decide what company you become.
It will simply execute the one you’ve already chosen.
Or the one you never did.