When to stop a marketing channel, decided in advance

Aug 22
3 min read

Updated: 2 days ago

Introduction

Businesses are good at starting channels and almost incapable of stopping them. Something gets launched, produces little, and stays funded because stopping feels like admitting waste.

So the budget spreads thinner every year across activities nobody would start today. The fix is not more discipline in the moment — it is deciding the exit before there is anything to defend.

1. When to stop a marketing channel is a question to answer before starting

The only honest time to set a stopping rule is before any money has been spent, because after that you are protecting a decision rather than evaluating one.

Write it down at launch: this channel gets this budget over this many weeks, and we continue only if it produces this result.

Two sentences, agreed in advance, remove almost all of the difficulty later. Everything below is about making those two sentences realistic.

2. Give it enough time to be judged, and no more

Every channel has a minimum honest test length, and it is longer than a fortnight.

The floor is however long your sales cycle takes plus enough volume to be more than noise. For an impulse purchase that might be three weeks; for a considered service, two months. Ten enquiries is not a result either way.

Set the window at launch. Extending it later because the numbers are disappointing is the same as having no window.

3. Judge on cost per customer, not cost per click

A channel that produces cheap attention and no customers is expensive.

Compare acquisition cost against what a customer is actually worth over their lifetime, not against a single transaction. A channel costing more per customer than the first sale can still be profitable if those customers return.

Without the lifetime number, every channel comparison is guesswork with decimal places.

4. Separate a broken setup from a bad fit

Before stopping, establish which of two very different things you are looking at.

A broken setup means the channel never got a fair run: conversion tracking absent, form failing on mobile, budget spread across nine campaigns, enquiries going unanswered for days. A bad fit means it worked mechanically and the audience did not buy.

Stopping a broken setup teaches you nothing and you will make the same mistake again. Fix the mechanics, restart the clock, then judge.

5. Look for the second-order value before cutting

Some channels underperform on direct attribution and still earn their place.

A channel that mostly generates branded searches, review volume or referrals will look weak in a last-click report. Check whether direct enquiries and branded search fell during any period the channel was paused — that is the cheapest available test of whether it contributes.

If you have never paused it, you do not know.

6. Reduce before you eliminate

Stopping does not have to be binary, and a step down is often the better first move.

Cut the budget to a third and hold it for a full test window. If results fall proportionally, the channel was working and you were buying volume. If they hold, you were overpaying for the same outcome.

That single experiment answers more than months of debate, and it is reversible.

7. Write down what you learned, then actually close it

A stopped channel that stays half-open is the worst outcome: no budget, no attention, still running.

Turn off the spend, cancel the tooling, remove it from the reporting, and tell whoever was running it. Then write two lines: what we tried, what it cost, what happened, and what would have to be different to try again.

Without that note the channel returns in eighteen months as a fresh idea.

8. Expect to be wrong sometimes

A channel stopped on evidence and later found to work is not a failure of the method.

Conditions change: platforms shift, competitors leave, your offer improves. The note from the last attempt is what makes a second attempt cheap, because you already know the setup and the cost.

Decide on the evidence you have, not on the possibility that different evidence exists.

Conclusion

Set the budget, the window and the required result before launching. Judge on cost per customer against lifetime value, and check whether a poor result reflects a broken setup rather than a bad audience.

Test the second-order contribution by pausing, step the budget down before cutting entirely, close it properly, and write the note that makes a future attempt cheap.