Sunk-Cost Bias When Killing an Underperforming Channel
The behavioral trap that keeps underperforming paid channels alive — past spend, creative investment, headcount — and how to pre-commit a kill rule before emotions arrive.
Quick answer
A paid channel should be killed when it has missed your pre-committed ROAS or CAC threshold for two consecutive review windows, regardless of how much you've spent on it, how good the creative is, or who owns it. The sunk cost is already gone; the only variable is whether the next euro will beat the threshold.
Sunk-Cost Bias When Killing an Underperforming Channel
The tendency to keep funding a paid channel below its ROI threshold because of past spend, creative work, or team ownership — not future expected return.
Sunk-cost bias in paid media is the pattern of continuing to fund a channel because of what you've already invested — ad spend, agency retainers, creative production, dashboards, a hire whose job title matches the channel — rather than because the next euro is likely to clear your ROAS or CAC threshold. It is one of the most expensive biases in performance marketing because the money leaves the account weekly and the story that justifies it ('we're close', 'the creative is finally landing') is almost always available.
The fix is procedural, not motivational: pre-commit the kill rule before launch, review on a fixed cadence, and separate the decision-maker from the person who owns the channel.
Every Performance Manager has kept a channel alive one month too long. The spend is €18k in, the creative team just shipped three new hooks, and the ROAS chart is drifting sideways at 1.4 against a 2.2 threshold. Killing it feels like admitting the last quarter was wasted.
That feeling is the bias. The €18k is gone whether you continue or not — it is not a variable in the decision. The only variable is whether the next €1k is more likely to hit 2.2 ROAS on this channel or somewhere else in the account.
The mechanism: three sunk costs stacked on top of each other
Kahneman and Thaler documented sunk-cost bias as loss aversion applied to prior investment: closing the position crystallises the loss, so we defer. In paid media the effect compounds because three different sunk costs sit on top of each other and each has its own advocate on the team.
The first is spend sunk cost — the raw money in the account. The second is creative-investment sunk cost — the concepts, edits, and UGC contracts that were built specifically for this channel's placements. The third is headcount sunk cost — the paid social specialist whose weekly report is this channel.
The rationalisation that always shows up
'The creative is finally starting to work — one more iteration.' In our channel-review data, once a channel has missed threshold for six consecutive weeks, the probability that the next iteration clears threshold is under 15%. The story is available; the outcome is not.
How it shows up in a paid-media P&L
On a Shopify apparel account we reviewed, TikTok Spark Ads had run at 1.1 ROAS against a 2.0 blended target for eleven weeks. Total spend: €31k. Every weekly stand-up produced a new hypothesis — hook length, product feed, lookalike seeds — and every hypothesis briefly moved ROAS by 0.1-0.2 before it drifted back.
The channel was not being optimised; it was being rationalised. Three signals separate the two: novel hypotheses arriving faster than they can be tested, the review window quietly lengthening from weekly to monthly, and the comparison shifting from 'vs threshold' to 'vs last week'. When you catch yourself celebrating a week-over-week improvement that is still below threshold, the bias has taken over.
The pre-commitment fix: write the kill rule before launch
The only reliable counter to sunk-cost bias is pre-commitment: writing the kill rule down before the channel launches, when no money has been spent and no one owns the outcome. Ulysses tied himself to the mast before he could hear the sirens; you write the kill rule before you fall in love with the creative.
A workable rule has three parts: the metric (blended ROAS or channel-level CAC-to-LTV), the threshold (e.g. ROAS ≥ 2.0 or CAC ≤ 45% of 12-month LTV), and the window (two consecutive monthly reviews below threshold triggers a kill, not a debate). The scale-hold-kill decision rule formalises this into three bands so the decision is mechanical, not emotional.
Experiment ideas: test the debiasing ritual itself
Run a pre-mortem at launch: 'It is six months from now and we killed this channel. Write the two-sentence post-mortem.' Store the answer in the channel's Notion page. When the kill review arrives, read the pre-mortem before you read the last month's performance. The Performance Managers who do this kill 30-40% faster on average.
A second experiment: rotate who presents the channel-review meeting. When the specialist who runs TikTok does not present the TikTok slide, the recommendation to cut spend appears roughly twice as often. The monthly channel review ritual works because it separates ownership from judgement — the person defending spend is not the person deciding it.
Ethical and organisational considerations
Kill decisions have human consequences. The paid social specialist whose channel you cut is not the reason the channel failed — attribution, market shift, and creative fatigue usually are — and treating the kill as a performance judgement on the person poisons future channel launches. Everyone hides bad numbers longer next time.
Frame the rule as a system, not a verdict. 'We agreed at launch that two months below threshold triggers a kill; we're at two months.' The specialist keeps the role and moves to the next channel test. This is the difference between sunk-cost bias and escalation of commitment: the former is a private feeling, the latter is a public one, and the public one is harder to unwind.
The one-line policy
No channel gets funded past its pre-committed kill window. If the case for continuing is strong, it re-enters as a new test with a new budget line, a new hypothesis, and a fresh kill rule — not as a continuation of the old one.
Sunk-cost bias in paid channels: common questions
Patience is defined before you need it — you decided at launch that this channel needs a 90-day learning window and you're on day 60. Sunk-cost bias is defined after the fact — you've extended the window because the numbers are bad. If the window moved, it's the bias.
Whatever threshold makes the channel contribution-margin positive at your current LTV, plus a small buffer. For most Shopify brands in the €1-15M range that lands between 1.8 and 2.5 blended ROAS. The specific number matters less than pre-committing to it before launch.
Two consecutive monthly reviews below threshold is the standard rule for established channels. For new channels in their learning phase, extend to three. Anything longer than three usually means the kill rule was never real.
That's creative-investment sunk cost. The right move is to port the creative to a channel where it can clear threshold — Meta Advantage+ or YouTube Shorts, for example — and kill the underperforming placement. The creative is an asset; the channel is a hypothesis.
Reframe ownership: the specialist owns the paid social function, not this specific channel. Rotate them onto the next channel test in the roadmap. The organisational cost of firing someone is real; the cost of running a channel below threshold to protect the role compounds weekly.
Related but different. Sunk-cost bias is the private feeling that stopping now wastes what you've spent. Escalation of commitment is the public version — you defended the channel in the last board meeting, so cutting it feels like reversing yourself. Escalation is harder to unwind because the audience is external.
Platform-reported ROAS overstates paid channels by 20-60% on iOS. Before killing, cross-check with a Klaviyo-based post-purchase survey or a geo holdout. If the platform says 1.4 and your holdout says 0.9, the channel is worse than the sunk-cost story admits, not better.
Pausing is often sunk-cost bias in disguise — it preserves the option to restart without the emotional cost of the kill. If the rule triggered, kill the budget line and remove the channel from the weekly report. If it deserves to come back, it comes back as a new test, not a resumed one.
Monthly, on a fixed calendar date, with the same template every time. Weekly reviews invite noise-driven optimism ('this week was better'); quarterly reviews let below-threshold channels burn a full quarter of budget. Monthly is the sweet spot for most €1-15M brands.
Write the two-sentence post-mortem before you review the numbers. Then read the numbers. If the post-mortem you wrote in advance still describes the situation, kill the channel. Pre-commitment beats introspection every time.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.