Writing

Who Owns the Number When the Funnel Breaks

Marketing and product argue about the same numbers because the funnel does not divide cleanly. It divides on control, not on stages, and that line can be drawn before anyone needs it.

A number misses. Marketing says the product is not converting what it sends. Product says the traffic is unqualified. Both have data, both are partly right, and the meeting resolves nothing because there is no agreed method for deciding.

This is usually treated as a people problem, solved with better relationships and more alignment meetings. It is not a people problem. It is a definitional one, and definitional problems are solvable.

The mistake is dividing the funnel by stage

The instinctive fix is to draw a line across the funnel. Everything above it belongs to marketing, everything below it belongs to product. Traffic and conversion here, activation and retention there.

It fails immediately, because most stages have mixed ownership. Take trial conversion, which looks like a clean marketing number. Whether a visitor signs up depends on the message they arrived with, the price they encountered, the length of the signup flow, and whether anything was broken that morning. That is marketing, leadership, product and engineering inside one metric.

So a line drawn through that stage produces a claim neither side can disprove. Both can point at real evidence, and the argument becomes a contest of conviction, which the more senior person wins.

Ownership follows control

Start instead from the only question that has a factual answer.

Who can change this without asking anyone, and see the number move?

That is the whole boundary. Everything else is derived from it, and it holds because accountability without control is not accountability, it is blame with a reporting line attached.

Marketing owns the promise. Who hears it, what it says, where it runs, and what expectation a person arrives with. Marketing can change every one of those unilaterally.

Product owns the delivery. What a person meets when they arrive, whether it works, and whether it does the thing. Marketing cannot change any of it, no matter how the quarter is going.

The boundary sits at the moment someone stops reading and starts using. Before that moment, the experience is made of claims. After it, the experience is made of software.

Four questions that assign any failure

With the principle fixed, a specific failure resolves by asking four questions in order and stopping at the first one that fails.

Did the right people arrive? Wrong audience, wrong channel, wrong intent. If the people showing up were never going to buy, nothing downstream matters. Marketing.

Did they arrive expecting the right thing? The right people came, and what they believed they were getting was inaccurate. Marketing owns the accuracy of the promise, not just its reach.

Did the thing they met work? Broken, slow, confusing, or missing entirely. No message survives a signup flow that fails on mobile. Product.

Did it deliver what was promised? Everything functioned and the buyer was still disappointed. This is the only genuinely contested question, and it is where every real argument between these two functions actually lives.

Resolving the fourth question

A gap between promise and delivery has exactly two causes, and they can be told apart.

Ask whether the product does what a comparable product in the category does. If it does, and buyers are still disappointed, the promise exceeded the category rather than the product falling short of it. Marketing overclaimed.

If it does not, the built product fell short of the intended one. Product underdelivered.

That test only works under one condition, and the condition is the whole mechanism: somebody has to have approved the claim before it shipped. Once product signs off on what marketing is going to say, product has accepted responsibility for meeting it, and marketing has accepted responsibility for not exceeding it. The gap becomes assignable because it carries a signature.

Without that step, the fourth question is unanswerable and always will be. With it, the answer takes about ten minutes.

Why the shortfall attaches to marketing by default

There is a structural reason these arguments tilt one way, and it has nothing to do with anyone acting in bad faith.

Marketing is measured on an outcome. Pipeline, revenue, qualified leads. Product is usually measured on activity: features shipped, velocity, roadmap progress. Those are not the same kind of number.

When a function with an outcome number and a function with an activity number disagree about a shortfall, the shortfall attaches to the outcome by default. Not because anyone decided it should, but because it is the only number in the room that visibly moved.

So the durable fix is not a sharper boundary. It is a second outcome number. Marketing owns qualified arrivals. Product owns converting them. Both appear on the same dashboard, at the same cadence, visible to the same people.

Now a degradation has two owners looking at the same screen, and neither can route around the conversation, because the number is public before the meeting starts.

What it costs to get this wrong

This can read as an internal fairness question, which makes it easy to defer. It is not. Misassigning a shortfall has a specific and expensive failure path, and it runs on a predictable schedule.

A number misses. The shortfall attaches to marketing, because marketing is the function carrying an outcome. The marketing leader is replaced. The new one arrives, spends a quarter learning the company, resets the strategy, and runs into the same constraint, because the constraint was never in marketing. Nine to twelve months later, the same conversation.

Two things happen in parallel, and the second is the one that ends companies.

The strategy never gets to run. Three leaders in three years is three resets, and nothing runs long enough to produce a readable result. So the company accumulates opinions about what does not work and no evidence about what does.

The actual constraint never receives a signal. This is the part worth sitting with. If activation is failing and marketing keeps getting replaced, the function that could fix activation is never told there is a problem. It is not avoiding accountability. It genuinely does not know, because every quarter the company concludes the issue was marketing and acts on that conclusion.

So the defect persists, protected by a diagnosis that keeps pointing somewhere else. Meanwhile a competitor who assigned it correctly in year one has spent three years fixing it.

Companies rarely fail from one bad decision. They fail from repeatedly solving the wrong problem while the runway shortens, and a misassigned number is one of the most reliable ways to do that. The revolving door is not the disease. It is the symptom that the disease has not been identified yet.

Agreeing this in advance

Everything above is straightforward to agree when nothing is wrong and nearly impossible to agree when something is. That asymmetry is the single most important thing to understand about this, and it means the work is calendar-driven rather than intellectual.

Do it at the start. Onboarding for a new marketing leader, or the first week of a quarter. A boundary drawn while the numbers are fine is a shared operating principle. The identical boundary proposed after a miss reads as an escape attempt, no matter who proposes it or how reasonable it is.

Write down the handoff event, not the stage. A stage is a concept and concepts are arguable. An event is logged. Agree the specific moment that divides the two, then let the instrumentation settle disputes instead of seniority.

Approve claims before they ship. A short review where whoever builds the thing signs off on what will be said about it. It takes minutes and it is the only thing that makes the fourth question answerable later.

Give both functions an outcome. If product will not accept a number, that is worth knowing on day one rather than in month nine, and it tells you something important about how the company actually operates.

Agree what an inconclusive test means. Sometimes marketing genuinely cannot vary the inputs enough to prove anything, usually because product controls the surface. Decide in advance that this counts as evidence rather than as an unresolved question. An inability to test is a finding about control, and control is what ownership follows.

What this is actually for

The point is not to win the argument. It is to stop having it.

Two functions that agree the boundary in advance spend their disagreements on what to do rather than on whose fault it is, and that is a materially different company from one where every miss produces a week of positioning. The method above takes an afternoon to agree and saves that week every time a number moves the wrong way.

None of it requires anyone to be generous or to concede anything. It requires only that ownership follow control, which is the one arrangement both sides can defend on their worst day.

Frequently asked questions

How do you decide whether a funnel problem belongs to marketing or product?

Ask who can change the variable without asking anyone and see the number move. Marketing controls who hears the promise, what it says and where it runs. Product controls what people meet, whether it works and what it does. Ownership follows control, because accountability without control is just blame.

Why does splitting the funnel by stage not work?

Because most stages have mixed ownership. Trial conversion depends on the message, the pricing, the signup flow and whether anything is broken, which sit with four different owners. Drawing a line through a stage produces an unfalsifiable claim that both sides can argue forever.

Where exactly does marketing's responsibility end?

At the moment someone stops reading and starts using. Whether a qualified person clicks to start is a function of who arrived and what they understood, both marketing variables. What happens after that click is the product surface, which marketing cannot change.

What are the four questions that assign a failure?

Did the right people arrive. Did they arrive expecting the right thing. Did the thing they met work. Did it deliver what was promised. The first two are marketing, the third is product, and the fourth belongs to whoever last approved the claim.

How do you settle a gap between what was promised and what was delivered?

Ask whether the product does what a peer product does. If it does and buyers are still disappointed, the promise exceeded the category and marketing overclaimed. If it does not, the built product fell short of the intended one. This is only answerable if someone approved the claim before it shipped.

Why does product need an outcome number too?

Because marketing is usually measured on an outcome while product is measured on activity such as shipped features and velocity. That asymmetry means every unexplained gap flows toward whoever has an outcome attached to their name, regardless of cause. Two outcome numbers on the same dashboard removes the escape route.

When should the ownership boundary be agreed?

Before the numbers are under pressure, ideally during onboarding or at the start of a quarter. A boundary drawn while everything is calm is a shared operating principle. The same boundary proposed after a miss is read as an attempt to escape accountability, whoever proposes it.

What if marketing cannot run a clean test because product controls the surface?

That itself is the finding. If marketing cannot vary the inputs enough to move the number, marketing does not have a lever there and cannot be accountable for it. An inability to test is evidence about ownership, not an obstacle to determining it.