How to Kill a Marketing Channel That Is Not Working

To kill a marketing channel, write the rule before you need it: trial length, outcome threshold, and the cost per outcome you won't accept. Then judge the channel on that agreed number, not on how you feel about it this week. Past its trial with no result, it goes today.

How to kill a marketing channel without second-guessing yourself

You already know which one it is.

There's an activity on your list that eats evenings and produces nothing you can point to. You keep it because it has a story. "It takes time." "That's where the audience is." "I've already put so much into it." You wrote that story. You're its advocate. And that's why you can't be trusted to judge it in the moment.

Most advice on marketing is about adding, not about how to kill a marketing channel. A new platform, a new format, a new funnel. The better move is usually subtraction. Finding the losers is a faster return than finding a winner, because every evening you stop wasting is an evening that goes straight into something already working.

It needs no new creativity and no new channel. It costs the willingness to look, and the nerve to act on what you see.

Why you can't cut by feel

Alone, decisions get settled by whichever thing you enjoyed most recently, or felt guiltiest about skipping. That isn't a character flaw. It's what happens in a vacuum. With no evidence, mood is the only currency.

So you keep things you like and abandon things you dread, and neither has anything to do with whether they produce customers. The reverse trap is just as real. You get a bad month on a channel, panic, and drop something that was about to pay. The channel wasn't failing. It was slow.

Both errors come from deciding in the moment. The fix is to decide earlier, when you're calm, and let the rule do the deciding later.

Free guide · 26 pagesStop paying for what does not work.The Attribution Playbook: how to find which marketing produces money.Get the free guide →

Kill a marketing channel by rule: four lines

A rule to kill a marketing channel fits on a card. It has four parts.

1. The trial length. How long the channel gets before you judge it. Set per channel, from its real lag.
2. The outcome threshold. What counts as working. Customers, or the furthest milestone that reliably predicts them. Not clicks.
3. The unacceptable cost. The most you'll spend, in evenings or money, for one outcome.
4. The decider. A sentence saying the call is made on the number, not on mood.

Write it while nothing is at stake. Put it where you'll see it on review day. The point is to take yourself out of the room at the moment you'd be most likely to negotiate.

An example rule, in words, for an archetype: say you run a productised audit and you've started posting on a new platform. "Twelve weeks. At least a handful of qualified conversations that lead to a paid audit. If one paid audit costs me more evenings than the audit earns, I stop. I decide on the tracker, not on how the posts feel." Yours will differ. The shape won't. Notice what the example doesn't say. It doesn't say "see how it goes".

Set the trial length from the lag, not from patience

This is where most people cut the wrong thing.

Some channels pay in days. Some pay in quarters. Content, a podcast, a community: they work long after the touch. Judge them monthly against something that pays back fast and they lose every time, and deserve to lose none of them.

So measure the lag. For customers attributed to a slow channel, how long was it between first touch and money? That tells you how long a fair trial actually is. Getting the window right matters here too, because a window shorter than your cycle erases the evidence. The method is in how to set an attribution lookback window.

Deciding the trial length before you start is the only reliable defence against cutting something in month four that pays back in month seven. Commit to it in writing. Then leave it alone until it's over.

If you don't know the lag yet, you don't have grounds to cut. You have grounds to measure.

Check these before you pull the trigger

Evidence that insults you is the entire point of the exercise. But a surprising number deserves one round of suspicion first. Not more than one.

Before a cut, run through this:

• Is the tagging right? A fragmented naming convention can hide a channel across three labels.
• Does the join hold? If forms drop the source, a channel can look dead when its leads simply lost their history.
• Is the window long enough? A short one erases slow channels.
• What does the other model say? A channel that looks weak on last touch may be your engine on first touch. See first touch vs last touch.
• What do people say? Check the self-reported answers. A channel nothing measures but everyone mentions has not been proven useless.

If it survives all five and still shows nothing, believe it. The whole point of building the instrument was to be able to be wrong cheaply. Refusing a finding turns the dashboard back into wallpaper.

Creators and harvesters: the cut that backfires

There's one cut that looks like discipline and isn't.

Line your channels up by first touch and by last touch. Some rank high on first and low on last. Those are your demand creators: content, a podcast, communities. Some rank high on last and low on first: branded search, retargeting, direct. Those are harvesters.

Harvesters look wonderful. Cheap, efficient, close to the money. And they can't manufacture demand, only capture it. Fund them by cutting the creators and you'll watch them starve over the following two quarters, wondering why the cheap channel stopped working.

Their efficiency was borrowed. Attribution is how you see the loan.

So the question isn't only "is this channel producing?" It's also "is something else living off this channel?" Check that before any cut. The reasoning behind the whole question is in how to know which marketing channel makes money.

Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →

What the decision looks like in practice

Most review days end in one of four outcomes. Having them named makes the call faster.

To kill a marketing channel well, the third row is the one that saves people. The fourth is the one nobody checks. A channel can produce volume and rate and still manufacture support work, which for one person is the worst outcome of all.

Do it today, and do it alone

Past its trial with no result, a channel is cut today. Not watched for another month. Watching is how a failed experiment becomes a permanent chore.

And change one significant thing at a time. If you cut a channel, double down on another, redo your landing page and change your pricing in the same month, the next month tells you nothing. You paid for four changes and learned zero. It feels slower to sequence them. It's much faster, because every change teaches you something.

Then write down the decision. A single paragraph: what you cut, what the number said, what you'll do with the freed evenings. A year of those notes is worth more than any dashboard, because it lets you check your own judgement and not only your channels. Next month, look at what the cut actually did before you make another. The routine is in how to review marketing results every month.

One more thing. Cutting isn't a verdict on you. The channel didn't fail you and you didn't fail it. It had a fair trial against a rule you wrote, and the evenings are worth more elsewhere.

Free guide · 26 pagesStop paying for what does not work.The Attribution Playbook: how to find which marketing produces money.Get the free guide →

If you've never cut anything

Here's a test. Has this ever made you stop doing something?

If not, it hasn't paid for itself. It's a comfort object, an expensive way of feeling informed. A system that never changes a decision is a cost.

Pick one channel today. Write the four lines. Put the trial end date in your calendar. That's the whole assignment, and it takes less time than the next post you were going to write for it.

Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →

Frequently asked questions

How long before I kill a marketing channel?

Long enough to cover its real lag from first touch to money, which you agree in advance. Fast channels get short trials, content and communities get longer ones. Decide the length before you start, so you don't relitigate it every month.

What if a channel produces traffic but no customers?

Traffic is not an outcome. Judge it on customers, or the furthest milestone that reliably predicts them. If a channel sends visitors but nobody pays, it's past its trial, and it passes your checks, it goes.

Should I cut a channel that looks weak on last touch?

Not on that alone. Check first touch and what customers say. A channel that creates demand for others to harvest looks weak at the close and strong at the start. Cutting it starves the channels that depend on it.

Is it a mistake to cut too early?

Yes, and so is cutting too late. A written trial length and threshold protects you from both. The rule decides, not your mood that week.

The full method, with the checklists, is in The Attribution Playbook, a free 26-page playbook.