Most workplace decisions come apart at the same hidden seam: a small, repeatable set of mental shortcuts that fires before anyone thinks to check for it. The hiring call, the budget cut, the “let’s keep going” on a project that’s clearly stalling. Different rooms, same handful of shortcuts, over and over.
In coaching conversations about a recent hiring or budget call, the same biases keep surfacing. Not exotic ones. Six of them, mostly, doing quiet damage while everyone in the room believes they’re being rational.
This piece is a mirror, not an encyclopedia. For each of six biases, you get a self-check (“you’re doing this when…”) tied to a specific work moment, plus one concrete counter-move you can run on your next real decision. If you want every documented bias in one place, the full glossary of decision-making traps and a wider catalog of biases in decision-making cover the encyclopedia version. This one is built to catch yourself in the act.
Why cognitive bias checklists rarely change decision-making
Most advice on bias hands you a list and assumes reading it is enough. It isn’t. Bias checklists treat decision-making like a knowledge problem. The real issue is timing. The shortcut fires automatically, and awareness shows up a beat too late to stop it.
Kahneman, Lovallo, and Sibony made this point in their HBR checklist for big decisions: the person making the call is the worst-placed to catch their own bias, because from the inside it’s invisible. Their fix skipped willpower entirely. It moved the check outside the individual and into the process, so the question gets asked whether or not anyone happens to feel biased that day. McKinsey’s work on the same problem, “Are you ready to decide?”, lands in the same place. The organizations that reliably debias their calls do it with structured routines built into the process.
Kahneman’s own book, Thinking, Fast and Slow, is blunt about the ceiling here. He spent decades studying these biases and still catches himself falling for them. That’s the reason a counter-move built into a routine beats a mental note every time: the routine runs even when your attention is somewhere else. So the goal below is narrow. For each bias, install one small check at the exact moment that shortcut tends to fire. That’s the whole game.
Anchoring: the first number in the room sets the frame
The first figure someone says out loud becomes the gravity well for the entire conversation. A candidate names 140k and every later number quietly orbits it. A vendor quotes twelve weeks and your plan reshapes around twelve without you noticing. You’re anchored when your final number lands close to the first one you heard, and you can’t reconstruct the math that supposedly got you there.
As an IC, you feel this when you’re scoping your own work. Someone drops “this should take about two days” in standup, and your estimate collapses toward two even though your gut said closer to five. The anchor did the negotiating for you.
As a manager, you set anchors without meaning to. Say “I’m thinking this role tops out around 120k” before your team weighs in, and you’ve capped the discussion before it started. Everyone now argues in the gravity of your number.
Counter-move: write down your own estimate before you hear anyone else’s. In hiring, set your range from the role and the market before the candidate names a figure. In planning, have each person put their number in writing privately, then compare. You want an independent anchor, a real second data point the room can genuinely argue against.
Confirmation bias: you already know how the debrief ends
You walk into the review having already decided the answer, and the meeting is theater. You notice the two data points that fit your read and skim past the three that don’t. You’re in it when you catch yourself thinking “I knew it” far more often than “huh, that’s strange.”
As an IC, this shows up the moment an A/B test result lands. You highlight the metric that supports the feature you built, and the flat conversion number gets a footnote and a shrug.
As a manager, you’ve quietly decided a report is struggling. Now every missed detail confirms the story, and every good week gets written off as a fluke, while the story about the struggling report stays intact.
Counter-move: before you look at the evidence, name out loud the one thing that would change your mind. “I’ll pass on this hire if the reference flags ownership.” “I’ll kill the feature if retention doesn’t move by X.” Writing the disconfirming condition in advance is what makes the difference, because it’s much harder to explain away a test you set for yourself when you were still honest.
Sunk cost: “we’ve already put six months in”
The strongest argument in the room for continuing is how much you’ve already spent. You hear “we can’t stop now, we’re too deep to turn back,” and everyone nods like it settles the matter. That sentence is the tell. Past investment can’t be recovered by spending more of it. The only question that actually matters is whether the next dollar, or the next week, is worth it on its own terms.
As an IC, you keep polishing a workflow you built because you gave it three weekends, even as the team quietly routes around it and uses something else.
As a manager, you defend a fading project to leadership using the headcount already burned, because killing it would mean admitting the original bet was wrong in front of the people who approved it.
Counter-move: ask the reset question. “If we were starting today, with zero already invested, would we choose to fund this?” If the answer is no, the six months is already spent and irrelevant to what happens next. This one stings precisely because it forces you to separate the decision from your ego and the story you’ve told about it.
Planning fallacy: your timeline is the best case in disguise
Your estimate assumes nothing goes wrong. No sick days, no scope creep, no dependency slipping a week. You’re in it when your timeline matches the exact version of events where every step goes perfectly, a version that has never once occurred in the history of your team.
As an IC, you commit to “done by Friday” while picturing three days of uninterrupted focus, forgetting the two review cycles and the one incident that always eats an afternoon.
As a manager, you roll up five optimistic estimates into a single launch date, then present it upward as a floor. It’s actually an optimistic ceiling that only holds if reality cooperates.
Counter-move: run a premortem. McKinsey’s writeup on premortems frames it cleanly. Before you commit, imagine it’s six months later and the project failed, then work backward through what went wrong. Assume the failure first, list the likely causes, then pad the plan against the two or three most probable ones. It surfaces the risks that optimism keeps hiding from you.
Authority bias: the room agrees the moment the senior person speaks
Opinions converge the instant the most senior voice states a preference. This is sometimes called the HiPPO effect, short for the highest paid person’s opinion. You’re watching it happen when a lively debate goes quiet and unanimous within thirty seconds of the VP saying what they think. The agreement looks like consensus. It’s usually deference wearing a consensus costume.
As an IC, you had a real objection to the architecture, then the principal engineer endorsed it, and you decided your concern probably wasn’t worth raising after all. The room never heard it.
As a manager, you state your lean first “to move things along,” then mistake the resulting nods for genuine buy-in. You walk out never learning what your team actually thought, which is the information you needed most.
Counter-move: collect opinions in writing before the senior person speaks. A quick round of written positions works well: everyone submits a sentence before discussion opens, or drops it in a Slack thread that closes before the meeting starts. Sequence matters too. Ask the most junior people first. The whole point is to get the independent read on the table before the anchor drops and the deferring begins.
Recency: the last thing that happened feels like the pattern
One strong week erases two mediocre quarters. One recent outage makes a stable system feel fragile. You’re in it when your judgment of a person, a vendor, or a project tracks the most recent event far more than the full record. The counter is a running note per person or project, so the record outlives your memory. Recency does enough damage in performance reviews to earn its own deep treatment. Why recency bias distorts reviews and how to stop it walks through the documentation cadence that actually fixes it.
What to do with a list like this
Reading six biases won’t debias you. That’s the honest limit. Kahneman studied these his whole career and still fell for them, and so will you. Awareness is simply too slow to stop a shortcut that fires on its own before you’ve had a conscious thought.
What awareness buys you is the ability to install the check. Pick your next real decision, a hire, a budget call, a go or no-go, and run only the two or three self-checks that fit it. Write your own number before you hear theirs. Name the evidence that would change your mind. Ask the reset question. The counter-moves work because they live in the process. Good intentions don’t survive a busy day.
If you want to go wider from here, the twelve decision-making styles smart managers switch between covers how to match your approach to the call in front of you, and the analysis-paralysis trap covers the opposite failure, when endless checking becomes an excuse to never actually decide. Both directions cost you.
Run this on one call and you’ll notice how often two or three of these fire at once, an anchor followed by confirmation followed by a sunk-cost defense of the whole thing. Catching even one of them changes the outcome.
The most reliable version of any of this is a check you don’t have to remember to run. That’s the real case for building it into a tool, or a teammate, that asks the question on cue. You can have Merlin surface a bias check before your next big call, so the reset question shows up when you actually need it, before the postmortem three weeks later.