“We’ve always spent there” is a lovely history lesson, but it’s not an investment case. Yet channels can stay in the budget long after the reason for funding them has disappeared. No one wants to be the person who cuts a source of demand, and the team can usually produce a plausible explanation for why the next month will be different.
A stopping decision does not require certainty that a channel has zero value. It requires a clear comparison between the best use of the next pound and the risk of withdrawing it. That comparison should include what the channel contributes, how strong the evidence is, what it costs to keep learning and what else the money could do.
Write the rule before the review
For each material channel, state the job it is supposed to perform. Is it expected to generate near-term contribution, reach a new audience, protect a source of demand, support a launch or create evidence for a later decision? Different jobs need different measures. A channel asked to build demand should not be judged only by last-click sales, while a campaign funded for direct acquisition should not escape a contribution test by claiming unspecified long-term influence.
Set a review date and the evidence needed by then. Specify the commercial outcome, time horizon, minimum data quality, relevant comparison and threshold for action. The threshold should reflect the business’s economics and the uncertainty it can afford, rather than a generic rule that one ROAS fits every channel.
Separate five possible decisions
Scale when the evidence supports additional investment and the business can serve the demand. Hold when the economics look acceptable but the evidence is too weak or immature to justify expansion. Fix when a specific, testable problem is limiting performance, such as a broken journey or poor product coverage. Test when uncertainty is material and a bounded experiment can reduce it. Stop when the best available evidence says the channel is unlikely to earn the next pound, or when the cost and risk of further learning exceed the potential value.
The difference between “fix” and “hold” matters. “Fix” needs a named cause, an owner and a deadline. Without those, the word can keep a weak channel alive indefinitely. “Test” needs a hypothesis and a result that could change the decision. Without those, it is another month of spend wearing a lab coat.
Include the alternative
Suppose a channel reports £60,000 in revenue against £15,000 in media. That 4x platform ROAS may look healthy, but the decision depends on margin, refunds, customers who would have purchased anyway and the contribution available from the next increment. If the best estimate of additional contribution before media is £11,000, continuing at the same level does not recover the £15,000 spend. If the estimate is uncertain, the next decision might be a capped test or a smaller hold, not automatic expansion.
Compare that option with the strongest alternative: another market, a product or checkout fix, customer retention, or simply preserving cash. Include any transition cost, contractual commitment and demand that may move elsewhere if the channel is paused. A channel can have real value and still lose the allocation decision because another use of the money is better.
Keep the door open only for a reason
A stop rule should state what would justify restarting. It might be a changed margin, a new product, stronger evidence of incremental demand or a measurement repair. This makes the decision reversible without leaving the budget half-open to habit. Keep a record of what evidence would change your mind, then review it when that evidence arrives.
A simple decision table can make the review concrete:
Scale: incremental contribution clears the agreed hurdle and capacity is available.
Hold: economics may work, but the evidence or customer horizon is incomplete.
Fix: a defined, repairable issue explains the shortfall.
Test: an unresolved assumption has a bounded test with a counterfactual.
Stop: the channel cannot justify the opportunity cost under the agreed rule.
There are no permanent decisions here. New evidence can move a channel from stop to test or from hold to scale. What shouldn’t move the decision by itself is the embarrassment of admitting that a previous allocation no longer earns its place. Good leaders own their mistakes and learn from them. The rest… don’t.
If your channel reviews keep ending with “let’s give it another month,” GV Marketing can help define the evidence and trade-offs that make the next decision clear.