Do You Have a Brand Problem or a Marketing Problem? A Four-Way Diagnostic

The request usually arrives fully formed. A new website. A rebrand. A campaign. Six months of content. By the time anybody outside the business is asked for a view, the shape of the answer has already been settled, and what remains is procurement: who does it, at what cost, by when.

That is not irrational behaviour. Discomfort is difficult to hold in its unnamed state, and a deliverable converts it into something with a start date and an owner. But a deliverable is an answer, and an answer produced before the question was stated is only accidentally right. What follows is a diagnostic rather than a recommendation: four problems that look nearly identical from inside a business, the signature each one leaves, and a test that separates them using evidence you already hold.

Watch how the choice actually gets made and you will notice that nobody proposes a diagnosis. Somebody proposes a thing. The thing comes from a small vocabulary the business already has: words it has heard from suppliers, read in a newsletter, or used the last time something felt wrong. A founder who has been through a rebrand reaches for a rebrand. A team with a marketer in it reaches for campaigns. A team with engineers reaches for a rebuild. The instrument you own shapes the injury you diagnose.

Underneath the proposal there is a diagnosis; it is simply unstated. The sentence we need a new website contains a claim that the site is the binding constraint, and that if the site were better the business would be better. That claim may be entirely true. It is worth noticing that it is a claim, because unstated claims cannot be argued with, and the brief that comes out of the meeting will be written to be commissionable rather than to be true. A commissionable brief has a scope, a budget and a date. A true one has a subject.

There is a second force at work. The person who names the problem is usually the person most exposed to it, and the name they reach for tends to sit just outside their own remit. That is not dishonesty, it is proximity. Whoever stands closest to a symptom sees the symptom rather than the constraint, and the constraint is frequently somewhere they have no authority to look. A diagnostic is worth running precisely because it does not care whose remit the answer lands in.

Four problems account for most of the unease a business feels about its own commercial performance. They are not the only four, and they overlap at the edges, but they fail in distinguishable ways. The useful question is not which of them exists, because several usually do. It is which one is the binding constraint: the one whose removal changes what everything else is measured against.

For each, three things are worth separating. What it feels like from inside the business. What it looks like in numbers you already hold, without commissioning anything. And the deliverable it is most often mistaken for, which is the most diagnostic of the three, because the misnaming is remarkably consistent.

A brand problem: what it looks like from inside the business

A brand problem is a problem of claim. People arrive, they grasp roughly what you sell, and they cannot work out what you are, who you are for, or why you rather than the alternative sitting next to you. The claim is either absent, or present but unowned, in the sense that it would be equally true of five other companies.

From inside, the tell is that your own people describe the business differently. Ask three colleagues what you do and for whom, and you get three answers that do not contradict each other but do not compose into one either. Sales conversations take a long time to reach the point where the buyer understands which category you are in. Proposals are compared against companies you do not consider comparable. The word basically does a great deal of work in your own explanations.

In the numbers you already hold, a brand problem sits at the point of understanding rather than the point of arrival. Traffic is rarely the complaint. People come, they read, they do not act, and when they do enquire the enquiry asks something your positioning should have answered before they reached the form. Referrals describe you inaccurately to the person being referred, which is the cleanest signal of all, because a referrer is motivated and still cannot repeat the claim.

It is most often mistaken for a design brief. A visual system is the part of brand work that is easiest to buy and easiest for a supplier to show, so it becomes the proxy for the whole. If the claim underneath is unresolved, a new visual system renders the old ambiguity more attractively and changes nothing about whether anyone can repeat it.

A demand problem: what it looks like in the numbers you already hold

A demand problem is a problem of distribution. The claim is clear and the offer is understood, and not enough of the right people ever encounter either. Everything downstream of arrival behaves acceptably, which is exactly what makes it difficult to see from inside: every conversation you have goes reasonably well, and there are simply not enough of them to matter.

The internal tell is that growth feels dependent on effort that does not compound. Revenue tracks how much outbound activity happened last month. Referrals and repeat work carry more of the total than anyone is comfortable saying out loud. Nobody can answer where the next ten enquiries would come from without naming a person rather than a channel.

In the numbers, a demand problem is legible at the top rather than at the bottom: few arrivals, thin discovery, a narrow set of entry points, and the sharpest signal of the set, enquiries that already know your name. If almost everyone who contacts you was told about you by somebody, you do not have a persuasion problem. You have a reach problem, and persuasion work will be applied to people who had already been persuaded before they arrived.

It is most often mistaken for a website problem, because the website is where you can watch people not converting, and it is the surface you control. Attention follows visibility rather than causation.

A product problem wearing a marketing brief

A product problem is a problem of the thing itself: what it does, who it genuinely suits, what it fails to do that the buyer assumed it would. It presents as a marketing brief more often than any of the other three, because its symptoms appear after the sale rather than before it, and post-sale symptoms are slower to be believed by the people who commission work.

From inside, the tell is churn that is explained rather than measured, and a steady pattern of concessions. Deals close on exceptions. The same objection is answered in every call by a person instead of by the product. Support handles an identical request repeatedly and it never reaches the roadmap, because it does not look like a defect. Somebody has started saying that it needs explaining better, about a thing the customer understood perfectly well and disliked.

In the numbers, a product problem shows as a gap between what people say at the point of purchase and what they do afterwards: strong intent, weak retention, usage concentrated in a fragment of what was built, expansion that never happens. The strongest signal is qualitative rather than numerical. When the reasons for leaving are specific and consistent, rather than vague and various, you are looking at the product.

It is most often mistaken for a demand brief. If the product is not being kept, more of it being bought does not accumulate. It drains at the same rate through a wider opening, and the cost of filling that opening rises every quarter it is not addressed.

A technology problem presenting as a conversion problem

A technology problem is a problem of mechanism: something does not work, or does not work for everybody, or does not work quickly enough to survive the patience available. It presents as a conversion problem for the simple reason that conversion is where it gets measured, and the measurement is taken at the far end of a chain nobody has inspected.

The internal tell is that failures are intermittent and therefore contested. Somebody says the form is broken, somebody else tries it and it works, and the matter is dropped. Enquiries arrive from one browser and not another, or arrive with fields empty. Nobody owns the estate as a whole, so nobody has checked the parts that only some people ever encounter.

This is the only one of the four with published criteria available to you, which makes it the cheapest to test and the least excusable to guess at. WCAG 2.2 sets out success criteria covering, among much else, whether a control can be operated by keyboard alone, whether the purpose of an input field is programmatically determinable, and whether an error is identified in text rather than by colour alone. A form that cannot be completed without a mouse fails a stated criterion. That is not a matter of taste, and it fails for every visitor who does not use one.

In the numbers, a technology problem is narrow and sharp: a drop concentrated at one step, on one device class, or under one condition. Breadth points away from it. A broad, even weakness spread across every step of a journey is far more often a claim problem wearing technical clothing.

It is most often mistaken for a copywriting brief, because rewriting a page is faster to arrange than instrumenting one.

Several of these problems usually exist at once. Only one of them is holding the ceiling in place.

Most wrong diagnoses collapse the moment somebody asks a second question. A few do not, and those are the expensive ones. What the durable misdiagnoses have in common is that the available evidence genuinely fits them, and that the correct diagnosis would require somebody in the room to accept a conclusion they are not positioned to accept.

• A demand problem read as a brand problem. The evidence fits: conversion is poor, the site does look dated, and the claim could certainly be sharper. It survives because sharpening a claim is defensible work that never visibly fails. What it misses is that the people arriving were never the right people, and a better claim aimed at the wrong audience converts no more of them than the old one did.
• A product problem read as a demand problem. This survives because the pipeline really is thin and the sales numbers really are real, so the remedy is aimed at something true. What it misses is why the pipeline is thin, which is what the previous cohort said after they left. The correct answer is uncomfortable in a way a campaign is not, because it lands on the people who built the thing rather than on the people who sell it.
• A technology problem read as a brand problem. It survives because a broken mechanism and a weak claim produce the same headline number when nobody has separated the steps. It is normally resolved by an afternoon of keyboard and cross-browser testing that nobody has budgeted for, which is precisely why nobody has done it.
• A brand problem read as a product problem. The most expensive of the set. Buyers hesitate, the team concludes the offer is insufficient, and features are added to close the perceived gap. The gap was one of comprehension, and every addition widens the surface that has to be explained, which makes the original problem worse in the name of solving it.
• A demand problem read as a technology problem. It survives because the instrumentation is honest and the drop-offs are real, and every funnel has drop-offs. Optimising a step that only a small number of people ever reach improves a percentage and moves no absolute number that anyone would notice.

The pattern underneath all five is that the surviving misdiagnosis is the one that stays inside the remit of whoever commissioned the work. That is worth stating plainly, and it is also worth resisting the cynical reading of it. People are not usually protecting themselves. They are reasoning from the evidence they can see, and the evidence they can see is bounded by what they are responsible for. The remedy is not suspicion. It is deliberately asking one question whose answer might belong to somebody else.

One question does most of the separating. Take each candidate problem in turn, assume it is completely solved overnight at no cost, then answer two things: who notices first, and what specifically changes in their behaviour. Not what improves. What somebody does differently.

The value sits in the second half. Conversion improves is not an answer, it is a restatement of the hope. The three people a week who currently ask what we actually do would stop asking, and would ask about scope and timing instead, is an answer, because it is falsifiable and because you would recognise it if it happened.

Work it through on a hypothetical business with a competent site, respectable traffic and too few enquiries. Suppose the claim were perfectly resolved overnight. Who notices? The people already on the site, at the moment they are deciding whether this is for them. What changes? Fewer of them abandon the page where you describe what you do, and enquiries begin arriving already knowing which of your services they want. Now suppose distribution were solved instead, and the claim left exactly as it is. Who notices? Nobody currently on the site, because they are already there. The count of arrivals changes and nothing about any individual visit does. Those two futures are different enough to be told apart from the outside, and that difference is what the diagnosis rests on.

Run the same test against the other two candidates. If the product were fixed, the people who notice first are existing customers, and the change is that they stop leaving for a reason you could already recite before you asked. If the mechanism were fixed, the people who notice are the ones who were already trying to act, because they had already decided to buy, and the change appears at one identifiable step rather than across the whole journey.

The identity of who notices first is the diagnosis. It is not a proxy for it. And if the honest answer to a candidate is that nobody would notice immediately, then whatever you have named is not the binding constraint, whatever else it may be. That verdict is not a failure of the test. It is the test doing its work before the money is spent.

The reason misdiagnosis is so durable is that it is not usually punished. Commission a rebrand for a demand problem and you will very probably receive a good rebrand. The discipline is real, the practitioners are skilled, and the work will be judged against the brief it was given rather than against the constraint it was supposed to remove. Everyone involved can behave competently and the commercial outcome can still be nothing.

The way deliverables are judged compounds this. A deliverable is assessed on craft, because craft is visible on the day it is handed over, while constraint removal is not visible for months. By the time the absence of effect is undeniable, enough has changed around it that attribution is hopeless. The natural next move at that point is to conclude that execution was insufficient and to buy more of the same discipline, which is how a misdiagnosis becomes a programme rather than a mistake.

The counter-argument deserves its due, because it is not weak. Work in the wrong discipline is not always wasted. A rebrand commissioned for the wrong reason forces a business to write down what it actually claims, and that act sometimes surfaces the real constraint more effectively than a diagnostic interview would have. A rebuild forces somebody to audit a system nobody had looked at in years. Serious work in any discipline generates information, and information is not nothing.

The limit of that defence is worth naming, though. It works by accident, at the full cost of the intervention, and it depends on somebody in the room being willing to abandon the brief when the evidence turns against it. If that person exists, say so at the outset and give them the authority explicitly, because otherwise the accidental discovery gets logged as scope creep and closed. If that person does not exist, do not count on the accident happening at all.

What would have to be true for this argument to be wrong? That the four disciplines are far less separable in practice than described, so that a claim, a channel, a product and a mechanism move together tightly enough that improving any one of them lifts the rest. That does genuinely happen in very small businesses, where one person makes every decision and an improvement propagates the same afternoon. The larger the organisation, the less it holds, because the four sit with different people and travel only as far as those people's remit reaches. If you are small enough that the objection applies to you, the diagnostic costs an afternoon anyway.

Often enough the diagnostic returns two answers and both are load-bearing: a claim that is unresolved and a mechanism that fails, or a product gap and a distribution gap feeding each other. When that happens, the interesting risk stops being the work and becomes the joins between the work.

Three things sit in the seam, and none of them belongs to a supplier by default.

• The decisions that fall between the workstreams. Whether the new claim changes what the product promises. Whether a technical constraint is worth designing around or worth removing. Whether a launch waits for the mechanism or ships without it. These belong to nobody unless one person inside the business is named to own them, and they are usually the decisions that set the ceiling for both workstreams at once.
• The form in which work passes across. A handover that arrives as a presentation is not a handover, it is a description of one. Agree before anything starts what each side hands the other, in what form, and what counts as complete: a written claim including the words that may not be altered, a specification with acceptance conditions, a content model with its fields and their meanings.
• Answerability for the whole. Each supplier will be answerable for their own part, and every part can succeed while the combined result underperforms. Somebody has to be answerable for the combination before it underperforms rather than appointed once it has, because an appointment made afterwards produces an explanation rather than a correction.

Whether that means one supplier or several is a genuinely open question, and it is not this article's to settle. It turns on how you evaluate judgement before you have seen any work, which is the subject of what to ask an agency before you hire one. What matters here is only that a cross-discipline diagnosis creates seams, and that the seams stay unowned until somebody is named to hold them.

The output of the diagnostic is a sentence, and the discipline of writing that sentence is most of the value. The form is fixed: a specific group of people, at a specific moment, does or does not do a specific thing, for a stated reason. Buyers who reach the services page cannot tell which of three offers applies to them, so they ask a question instead of making an enquiry. Nobody outside a referral ever arrives at all. Customers renew once and not twice, and say the same thing on the way out.

Written that way, the sentence carries its own test. Every element in it can be checked against something, and if it turns out to be wrong you find out early and cheaply, before a brief has been built on top of it. A sentence that cannot be wrong is not a diagnosis, it is a mood.

Then the sentence chooses the work, rather than the work choosing what the sentence is allowed to say. A brand verdict raises a further question this piece does not settle, which is how much of the identity actually has to change and what is being bought when it does: that is the subject of what gets sold when a brand stops working. A demand verdict points at distribution and at which entry points exist, rather than at the page the people who already found you are reading. A product verdict points at what the last cohort said on the way out. A technology verdict points at one step, one device class, one condition, and it is testable this week against criteria that are published rather than invented.

If the sentence will not resolve, and two candidates remain equally plausible after the counterfactual test, that is a legitimate outcome rather than a failure. It names its own next step, which is to instrument the thing you cannot currently see, rather than to commission work against a guess and hope the guess was a generous one.

One thing is worth saying about incentives before you take the diagnosis to anybody. A supplier who sells only one of the four disciplines will find one of the four, and will usually do so honestly. Zealsync runs brand, growth and technology work through Intense Path as a single practice for that reason, and the diagnostic above is deliberately written to be runnable without anybody's help. If you would rather work through it with somebody than alone, bring the problem rather than the deliverable.