How Long Should You Give a Channel Before You Kill It?

How long should you give a channel before you kill it?

Long enough to see one full sales cycle plus the time it takes the channel to warm up, and no longer. For most B2B businesses that is one to two quarters, decided before you spend anything. The date matters more than the duration, because a decision date is what stops a channel dying slowly.

The reason teams get this wrong is not impatience. It is that nobody wrote down what would count as working, so every review becomes a debate about whether to give it more time, and more time always wins until the budget runs out.

So the fix is upstream. You decide the window and the evidence before the first pound is spent, and then you hold yourself to it.

Why does the sales cycle set the floor?

Because a channel cannot show you revenue faster than your business can close a deal. If your average cycle is ninety days, a sixty day test cannot possibly tell you whether the channel produces customers. It can only tell you whether it produces conversations.

That distinction is the whole game. A short test measures the top of the funnel. A long test measures the bottom. Deciding which question you are asking tells you how long to wait.

If you do not know your cycle length, that is the first thing to fix, and it is a better use of a week than any channel experiment. You cannot evaluate anything against a number you do not have. It is the same problem as reading CAC payback when you sell services, where the denominator has to exist first.

What should you measure before revenue arrives?

Leading indicators that sit on the path to revenue. Qualified conversations, not clicks. Replies from the right job titles, not impressions. Demo requests from companies that look like your customers, not sign ups from anyone.

The test for a good leading indicator is whether it has ever preceded a closed deal for you. If it has not, it is an activity metric and it will comfort you rather than inform you.

Pick two, write them down with a number beside each, and check them at the midpoint of your window. The midpoint check is not a decision point. It is a chance to fix execution before the real decision arrives.

How do you separate a bad channel from bad execution?

Ask whether comparable businesses reach your buyer there. If they do and you are not, the channel works and your execution does not. If nobody like you is reaching anyone like your customer there, the channel is wrong for you and better execution will not save it.

This is the single most useful question in the whole exercise, and it is answerable in an afternoon of looking rather than months of spending.

When it is execution, the usual culprits are the offer and the targeting rather than the creative. Most channels I have seen written off had a message that did not match what the audience was there to do.

What does a fair test actually require?

Enough budget or effort to produce a readable signal, one message held constant long enough to learn from it, and a landing experience that matches the promise. Change all three at once and you will learn nothing from any of them.

Underfunding is the quiet killer. A test too small to generate meaningful volume returns noise, and noise gets interpreted as failure. It is better to test one channel properly than three badly.

Consistency matters just as much for organic channels. Publishing for six weeks and stopping tells you nothing about content as a channel, because the mechanism you were testing needs accumulation to work at all.

When should you kill it early?

When the audience is provably wrong, when the unit economics cannot work at any conversion rate you could plausibly reach, or when the work is unsustainable for whoever is doing it. Those three are worth acting on before the window closes.

The economics case is arithmetic you can do in advance. If the cost of a qualified conversation exceeds what a closed deal is worth to you, no amount of optimisation rescues it, and continuing is a decision to lose money slowly.

The sustainability case is underrated. A channel that only works when the founder posts daily is not a channel, it is a personal obligation, and it will end the first busy month. That is a real finding, not a failure of discipline.

When should you extend the window instead?

When the leading indicators are moving in the right direction but have not reached the level you set, and you can name what changed. Extending because you feel hopeful is how budgets disappear. Extending because replies doubled after you fixed the offer is a decision.

Write the extension down with a new date and a new threshold. An extension without a fresh commitment is just a channel that never dies.

Limit yourself to one extension. If the second one is tempting, the honest answer is usually that you are unwilling to accept the result, and that is worth admitting out loud.

Is killing the same as stopping forever?

No, and treating it that way makes the decision harder than it needs to be. Most channels should be paused rather than killed, with a note about what you learned and what would need to be true for you to try again.

Conditions change. A channel that failed when you had no proof can work once you have customer stories, and one that failed at your old price may work at a new one. The note is what lets future you reuse this quarter's expensive learning.

What should die permanently is the version of the experiment that was run badly. Write down why it was bad, so nobody repeats it in eighteen months and calls it a fresh test. The decision about what to try next is the one I covered in when to add a second channel and when to wait.

What do you do with the budget you free up?

Put it into the channel that is already working, before you try a new one. Doubling a working channel is almost always a better bet than starting from zero somewhere else, and it is the option teams skip because it feels less interesting.

Only when the working channel is genuinely saturated does a new experiment make sense, and saturation is rarer than founders think. Most businesses stop growing a channel long before it stops responding.

When you do start the next test, run it the same way. A date, two leading indicators, one message, enough budget to read the result. The method is what compounds across experiments, not any individual channel.

What should you do next?

Take the channel you are currently unsure about and write one sentence: by this date, we need this many qualified conversations, or we stop. If you cannot fill in the numbers, that is the reason it has survived this long without a decision.

Then check whether you are measuring something that has ever preceded a real deal. If not, change the metric before you change the channel, because you may be about to kill something that was working in a way you were not looking at.

If you want a second opinion on whether a channel is failing or merely early, reach out. It is a short conversation that saves a quarter of spend more often than not. Let's chat.