Killing Channels That Aren't Working (And Why Teams Struggle To)

Every growth team we work with has at least one channel, tactic, or programme that's running well past the point where it's demonstrated that it doesn't work. The team knows it's not producing returns. Leadership knows. Sometimes even the agency running it knows. And yet it continues, quarter after quarter, consuming budget and attention that would produce more elsewhere. The inability to kill things is one of the most consistent patterns in B2B growth operations, and one of the most expensive.

This article is about why teams struggle to kill channels, the specific cost of not killing them, and the operational discipline that makes clean kills possible. It's the hardest part of the optimisation pillar to get right because it requires admitting something didn't work, which teams are structurally bad at.

Why It's Hard To Kill Things

Four reasons channels persist longer than they should. Each one is operationally real and each one has to be addressed to produce a cleaner kill discipline.

Sunk cost reasoning. The team has invested in the channel. Tools. Creative. Optimisation time. Agency relationships. The mental arithmetic treats that investment as something to protect rather than a cost that's already paid. "We've spent six months building this out; we can't just stop now." In fact, the six months is gone regardless, the only question is whether the next six months produce return or waste. Sunk cost reasoning systematically blocks the right answer.

Attribution ambiguity. Most B2B companies don't have clean enough attribution to be confident about which channels are producing pipeline. This creates a permanent excuse to keep channels running: "we don't really know it's not contributing." The uncertainty is used to defer decisions that would be more clearly made if the data were sharper. In practice, for most B2B attribution questions, the data is sharp enough, the ambiguity is often a rationalisation rather than a genuine measurement problem.

Personal or political attachment. Someone on the team championed this channel. Someone built their plan around it. Someone's quarterly performance depends on it. Killing it creates internal friction, the champion feels exposed, the quarter's plan needs rewriting, relationships need renegotiating. The soft cost of internal friction often outweighs the hard cost of continued waste in how the decision gets weighed.

Fear of permanent closure. Teams worry that pausing a channel means abandoning it forever, and they'd rather keep something running than lose the option to use it later. In reality, pausing a channel for six months and returning later with fresh learning almost always outperforms running it continuously at marginal effectiveness. But the psychological difference between "pause" and "kill" matters, and teams conflate the two in ways that prevent both.

What It Costs To Not Kill

The case for kill discipline is usually framed in terms of wasted spend. Budget going to a channel that isn't producing. That's real, but it's the smaller of two costs.

The larger cost is attention displacement. A channel that's running, even poorly, consumes team bandwidth. Creative gets commissioned for it. Reviews include it. Optimisation cycles touch it. The team member responsible for it spends hours on it that could be spent elsewhere. A channel with £10k/month spend that isn't producing also costs 10-20% of someone's time, which at Series A team sizes is genuinely scarce.

The hidden cost is decision quality degradation. Teams running too many marginal channels develop a habit of not killing things. That habit spreads. Once the team is comfortable leaving underperforming paid media running, it becomes easier to leave underperforming content programmes running, underperforming sales sequences running, underperforming product features running. The kill muscle atrophies, and decision quality drops across the operation.

Teams that develop a clean kill discipline usually find that the second and third rounds of killing are dramatically easier than the first. The first kill is culturally awkward. By the third, the team treats it as routine operating practice.

The Kill Decision Framework

Four questions that, asked honestly, produce clear kill decisions most of the time.

1. What was the channel supposed to produce by now?

Go back to when the channel was launched. What was the expected outcome at the 6-month mark, the 12-month mark? If no specific expectation was set, the channel was never properly scoped and the first discipline is retroactively setting one: given reasonable expectations for this kind of channel at this spend level, what should have happened by now?

This question often surfaces the real diagnostic. "It was supposed to produce £50k in attributable pipeline by month 6; it's produced £8k by month 9" is a clear kill candidate. "We don't really know what it was supposed to produce" means the channel has been running without a goalpost, which is itself a reason to reset.

2. Is the gap explainable by external factors, or is it a channel problem?

Sometimes channels underperform because something changed, algorithm shifts, competitive entry, audience fatigue from market saturation. These external factors might mean the channel is still fundamentally viable but needs repositioning. Sometimes the underperformance is internal, the channel just doesn't work for this ICP at this scale, and no amount of optimisation will change that.

Honest diagnosis: can you point to specific external factors that plausibly explain the gap? If yes, consider repositioning rather than killing. If no external factors account for it and you've already done the reasonable optimisation, the channel is probably structurally wrong, not tactically fixable.

3. What specifically would have to happen for this to start working?

A clear answer to this question means the channel has a path. "If we shift targeting to mid-market from enterprise and cut CPL target by 20%, we think it would convert" is a path, now you can decide whether the path is worth pursuing. "If we just keep going and optimise harder" is not a path; it's continuation without a thesis.

Channels that have specific, testable paths to viability can be given one more quarter to prove or disprove. Channels with no specific path should be killed, because continuation without a thesis is just hoping.

4. What would the team do with the reclaimed capacity?

This is the question that makes the decision commercially real. If you killed this channel tomorrow, what would you do with the budget and the team time? If the answer is "shift to a known-working channel and expand it," killing is almost certainly correct. If the answer is "we'd have to figure out what to do with the budget," the channel might be worth preserving as a capacity placeholder even if it's not producing pipeline, because the alternative uses aren't clear.

In practice, most teams have obvious alternative uses for reclaimed capacity and aren't honest about them because acknowledging them would force the kill decision.

The Kill Mechanics

Once the decision is made, the operational kill has specific steps that produce clean exits rather than messy ones.

Announce the kill explicitly. "We're pausing LinkedIn paid effective end of month." Not a quiet de-emphasis; an explicit decision communicated to the team. This matters because implicit kills don't reclaim capacity, the team members involved keep doing marginal work on the channel until someone tells them not to.

Set a review date, not an end date. "We're pausing for six months; we'll review whether to restart in May." This addresses the fear-of-permanent-closure objection. The channel isn't gone forever; it's on hold, with a specific date when the decision gets reopened.

Document the current state. What was running, what was learned, what might work differently if the channel is reactivated. This matters for two reasons: it makes future reactivation cheaper, and it captures the institutional knowledge that would otherwise be lost if team members change.

Reallocate capacity explicitly. The reclaimed budget goes to a specific named use. The reclaimed team time goes to a specific named project. Without explicit reallocation, reclaimed capacity dissipates without producing value, it just gets absorbed into general overhead.

Hold the kill. The test of whether the kill is real is whether, three weeks in, someone suggests restarting the channel because of a specific opportunity and the team says no. Kills that unravel within a month were never really kills.

When To Reconsider A Killed Channel

Channel kills aren't permanent, and reactivation can be legitimate. Three conditions that warrant reopening the decision.

Something external changed. The channel's platform shifted meaningfully (new ad format, audience expansion, pricing shift). Your market changed (category is suddenly active, competitor dynamics shifted). An external change creates a new context in which the old reasoning may not apply.

Your own positioning or offer changed. An ICP rewrite, a new product, a fundamental commercial model shift, any of these might make a previously-killed channel newly viable. The channel wasn't wrong in the abstract; it was wrong for the previous commercial context.

Capacity and budget have expanded dramatically. A channel that wasn't viable at £5k/month might be viable at £50k/month because volume unlocks different tactics. Pausing at Series A and reopening at Series B is a legitimate pattern.

Outside these conditions, the original kill logic still applies. Teams that reopen channels without a specific triggering reason usually end up back in the same state six months later.

Where Kill Discipline Sits in the NBT Growth Model

Channel kill decisions sit inside Stage 4, OKR-Driven Optimisation. They're the specific practice that makes the Optimisation stage meaningful: a stage that can measure but can't kill what isn't working produces continuous measurement without improvement.

The relationship to test-and-learn that compounds is direct. A compounding test-and-learn system depends on willingness to act on what tests reveal. Tests that reveal a channel isn't working, followed by continued investment in the channel, break the feedback loop that makes compounding possible.

The relationship to multi-channel paid plans is also direct. That piece argues for sequential channel addition, adding the second channel only when the first is proven. This piece is the other side of that coin: killing channels that have proven themselves not to work, rather than continuing them out of sunk-cost reasoning. Together they form a channel portfolio discipline that compounds rather than accumulates.

If your team is running channels that have demonstrated they don't work but continue anyway, or you want help making the kill decision on a specific programme, get in touch for a strategy call. Sometimes an outside voice is what unblocks the decision the team has already made internally but can't commit to.