Sunk-Cost Bias When Reading Marketing ROI
Sunk-cost bias is the operator trap of protecting a paid channel because of what you've already spent learning it. Here's how to read the ROI calculator without anchoring on history.
Quick answer
Historical spend is not a reason to keep a channel alive. If forward-looking ROI on your Marketing ROI Calculator sits below your minimum scaling threshold for two consecutive decision windows, cut or pause — the money already spent is gone either way.
Sunk-Cost Bias When Reading Marketing ROI
The tendency to keep funding an underperforming paid channel because of prior investment, not because of expected future return.
Sunk-cost bias in marketing ROI shows up when an operator looks at a channel's calculator output, sees negative or sub-threshold return, and still refuses to cut it — reasoning that "we've already spent €40k learning this channel" or "the creative team just shipped a new batch." That prior spend is unrecoverable and irrelevant to the next euro's return. The trap is emotional, not analytical: the ROI number is telling you one thing, and your ownership of the history is telling you another. Operationally, it inflates blended CAC, delays reallocation, and quietly starves the channels that actually work.
The bias is strongest right after a big-ticket investment: a new agency retainer, a shot-in-studio creative package, or a six-month attribution rebuild. The larger the sunk amount, the more the brain wants to justify it.
This page is about reading the output of the Marketing ROI Calculator honestly — separating what a channel has cost you from what it will return next month.
Why the bias happens
Two mechanisms drive it. First, loss aversion: cutting a channel feels like realising a loss, whereas continuing feels like preserving optionality — even when the expected value is worse.
Second, identity and effort justification. If you personally pushed to launch TikTok Ads six months ago, killing it reads as admitting the initial call was wrong. Operators unconsciously trade a small ego cost today for a larger P&L cost over the next quarter.
The tell
If your justification for keeping a channel contains the phrase "we've already invested" or "we just need a bit more time to make it work," you are reasoning from sunk cost. A healthy justification only references forward numbers: expected CPA, expected volume, expected incrementality.
How to detect it in your own decisions
Look at the channel's rolling 30-day ROI on the calculator, not the lifetime ROI. Lifetime numbers embed the sunk spend and blur the current picture — a channel that was profitable in Q1 and bleeding since April can still show positive lifetime ROI while burning cash today.
The second signal: run the "stranger test." If a new Head of E-commerce joined tomorrow, saw only the last 30 days of channel data, and had no history with your creative team — would they keep the channel on? If the answer is no, you're holding it for sunk-cost reasons.
How to fix it: a two-window kill rule
Pre-commit a decision rule before you look at the numbers. A workable default: define a decision window (typically 14 or 28 days for a paid social channel), set a minimum ROI threshold, and if the channel misses it two windows in a row with no obvious external cause, pause spend.
Pre-committing matters because the rule is easier to write when no money is on the line. Once you're €40k in, every threshold feels arbitrary and every exception feels justified. The minimum ROI to scale a paid channel is a good anchor for what "healthy" looks like — use it as the floor, not the ceiling.
Experiment ideas to de-bias the read
Run a forced 14-day pause on any channel that's been sub-threshold for a month. If blended ROAS stays flat or improves during the pause, the channel wasn't contributing incrementality — it was cannibalising demand from cheaper sources. A Shopify apparel brand doing this on branded search often finds Meta was riding on top of organic intent.
Second experiment: a blind review. Strip the channel names from your calculator output, share the anonymised table with a peer, and ask which they'd cut. When TikTok is labelled "Channel C," the sunk-cost anchor disappears and the decision usually gets sharper.
Ethical and organisational considerations
Killing a channel has human consequences — an agency contract, an in-house specialist's roadmap, a creative team's shipped work. Acknowledge them, but keep them out of the ROI read itself. Handle the people question in the follow-up conversation, not inside the spreadsheet.
Culturally, teams that celebrate "clean kills" as much as launches are far less prone to sunk-cost drag. Make it explicit that pausing a channel after a fair test is a good outcome, not a failure.
One-line reframe
The right question is never "was this spend worth it?" — it's "if I had this money in cash right now, would I put it here?" If no, reallocate.
Frequently asked questions
A minimum of two full decision windows — typically 4-8 weeks for Meta or TikTok, longer for lower-volume channels. Anything shorter conflates learning-phase noise with true performance. Pre-commit the window before you launch so sunk-cost bias can't stretch it later.
Yes — but "strategic" needs a forward-looking definition, not a historical one. If a channel is unprofitable today but you can articulate the specific mechanism (audience-building for a launch, learning phase on a new creative format) and set a checkpoint, that's strategy. If your reason is "we've invested so much," that's sunk cost.
The minimum ROI to scale a paid channel is your objective floor — it should be set based on contribution margin and target payback, independent of any specific channel's history. Sunk-cost bias tempts you to quietly lower the threshold for channels you've already invested in. Don't.
Reframe the conversation around expected forward return per euro, not lifetime ROI. Show two calculator outputs side by side: the historical view and the trailing 30-day view. When the delta is stark, most CFOs pivot quickly — they're often the easiest audience for a clean forward-looking argument.
Yes, though the shape is different. On SEO, sunk cost shows up as continuing to invest in content clusters that never ranked; on email, as protecting a segment that no longer converts. The kill rule is the same: forward expected return per hour or per euro of next investment.
A hard pause gives you cleaner incrementality data — you learn whether the channel was actually contributing. Ramp-down keeps some signal alive but muddies the read. Pause for at least one full sales cycle if you genuinely want to know.
Write the relaunch conditions at the moment you kill it: what specifically would need to change (new creative angle, new audience, platform-level pricing shift) before you'd try again. Without pre-committed conditions, you'll re-launch on emotion within a quarter.
It isolates forward-looking ROI on a per-channel basis so you're not reading a blended lifetime number that hides the recent trend. Filtering to a rolling 30- or 60-day view strips out the historical spend that anchors your decision-making.
Agencies are structurally sunk-cost-prone — they've invested account time you don't see. Ask them to submit a forward-looking plan with expected CPA and volume for the next 30 days, and hold the decision against that. If their plan matches your threshold, keep going; if not, the answer is clear.
Rarely, but yes: when the channel provides measurable brand or retention value that isn't captured in first-order ROI, or during a defined test-and-learn phase with a hard end date. Both require an explicit written justification that doesn't mention prior spend.
Test ideas before you ship them
Run unlimited A/B tests, attach hypotheses to outcomes, and build a searchable archive of what works — and what doesn't.