A company spent nearly a million dollars on failing software and chose to continue. Not because the future looked promising — because the past felt too heavy to abandon.
Cosmos Data Tech published a case study that should be required reading for every technology leader: a company spent nearly a million dollars and hundreds of labor hours on software that was, by everyone's private admission, failing. When the team sat down to decide whether to continue or switch, they chose to continue. Not because the future looked promising — because the past felt too heavy to abandon.
This is the sunk cost fallacy, and it is running more technology stacks than any architectural decision ever made. SUE Behavioural Design described it in February 2026 as one of the most expensive cognitive biases precisely because "it disguises itself as loyalty, perseverance, and a sense of responsibility." It feels like commitment. It looks like good stewardship. And it costs more every quarter you don't name it.
The Standish Group's CHAOS Reports have tracked IT project outcomes for three decades. Their baseline finding remains largely unchanged: only 16.2% of IT projects complete on time, on budget, with full features. The majority either fail outright or deliver partial value at inflated cost. And yet, the sticker price remains the smallest line item in most technology adoption decisions — because nobody calculates the ongoing cost of staying.
The question that breaks through the sunk cost fallacy is simple: "What is it costing us to keep using this?" Not what we've already spent — that money is gone regardless. What are we spending every month in workarounds, manual processes, lost productivity, and frustration? That number is almost always larger than the switching cost, and it's almost never calculated.
This is why the problem usually isn't you — it's the system you inherited and the psychological weight that prevents anyone from questioning it.
Daniel Kahneman and Amos Tversky's Prospect Theory established that people feel losses roughly twice as intensely as equivalent gains. In technology decisions, this means the perceived loss of admitting a tool doesn't work is psychologically heavier than the measurable gain of replacing it with something that does.
I wrote about this pattern in how cognitive bias shapes technology decisions — loss aversion is the single most predictive bias in enterprise technology adoption. But the sunk cost version is more insidious because it operates on accumulated loss. Every month you've paid, every integration you've built, every workaround your team has memorized becomes another brick in the wall between you and the obvious decision.
Forty-two percent of companies moved back to monoliths or modular monoliths in 2026 after premature microservices adoption. That statistic represents thousands of teams that eventually did the math — but only after years of technology fragmentation that could have been avoided by asking the question earlier.
The sunk cost fallacy has a vocabulary. Once you recognize it, you hear it everywhere:
"We've already invested too much to switch." The investment is gone. The only question is whether the next dollar goes toward something that works or something that doesn't.
"Our team knows how to use it, even if it's clunky." Expertise in a broken system is not an asset — it's a switching cost disguised as institutional knowledge. Your current tech stack should inform decisions, not prevent them.
"We can't justify the cost of switching after all we've invested." You can't justify the cost of staying. Run the numbers on the workarounds alone — the hours spent exporting data manually, the integrations held together with scripts nobody documented, the vendor sprawl that grew because the core system couldn't do what it promised.
SUE Behavioural Design recommends designing "psychological permission to abandon investments." In practice, this means creating structured moments where the question isn't "should we continue?" but "if we were starting today, would we choose this?"
The zero-based evaluation is the most effective tool. Strip away the history. Pretend you're evaluating this vendor for the first time. Would you sign the contract today, knowing what you know? If the answer is no, the only thing keeping you is the fallacy.
The second intervention is separating the decision-maker from the original advocate. The person who chose the platform has the most psychological investment in continuing. They need to be in the room — their context is valuable — but they shouldn't have veto power over the exit decision. Default bias compounds sunk cost: the existing system is the default, and defaults persist unless someone actively overrides them.
Every dollar and hour spent maintaining a system you know doesn't work is a dollar and hour not spent on the system that would. The sunk cost fallacy hides the opportunity cost by keeping your attention on what you've already spent instead of what you're currently losing.
The businesses that build efficiently aren't the ones that never make bad technology choices — they're the ones that recognize bad choices faster and have the discipline to act. That's not disloyalty. That's operational maturity.
Listen for the language: "We've already invested too much," "Our team knows how to use it," or "We can't justify the switching cost." If the primary argument for keeping a system is what you've already spent rather than what it will deliver going forward, the sunk cost fallacy is operating. Run a zero-based evaluation: would you choose this system if you were starting today?
Switching costs are real but almost always overestimated relative to the cost of staying. Calculate the monthly cost of workarounds, manual processes, and lost productivity on the current system. Multiply by 12-18 months (typical migration timeline). If the staying cost exceeds the switching cost — and it usually does — the math supports moving.
Because the fallacy operates through legitimate-feeling emotions: responsibility, consistency, and waste aversion. Kahneman and Tversky showed that losses feel twice as intense as gains. Abandoning a platform feels like accepting a loss, while continuing feels like perseverance. The Standish Group data shows that only 16.2% of IT projects fully succeed — most technology bets don't pay off, and recognizing that faster is a competitive advantage.
Yes, for context — no, for veto power. The original decision-maker has the deepest understanding of why the platform was chosen and what it was supposed to solve. That context is valuable. But they also have the strongest psychological investment in justifying the original choice. Include them in the evaluation. Don't give them the final call on the exit.
AI doesn't just reflect training data bias — it amplifies the biases of the people using it. Five cognitive shortcuts that derail technology decisions, and the structured processes that counteract them.
NLP is scientifically contested and commercially overhyped. I've studied it anyway, for years, because the pattern-recognition framework changed how I read every business conversation.
I have credentials in marketing psychology, consumer behavior, and positive psychology. Not because I wanted to be a therapist — because I wanted to understand why people do what they do.
Work With Us
Kief Studio builds, protects, automates, and supports full-stack systems for businesses up to $50M ARR.
Newsletter
Strategy, psychology, AI adoption, and the patterns that actually compound. No spam, easy to leave.
Subscribe
