Anchoring Bias: A 6-Question Reset Before You Decide

Early in my design career, a client opened a project call with a sentence I can still hear: "Our budget is around three thousand." I had planned to quote five. For the rest of the call I negotiated quietly with myself, trimming scope, discounting hours, until the number I finally said out loud sat just above theirs. Walking home, I realised the strange part. Three thousand was not a fact about the work. It was a fact about the conversation. Yet it had redrawn my entire sense of what the work was worth.

Psychologists call this anchoring: the first number, option, or opinion that enters a decision pulls every judgment after it toward itself. The classic evidence is almost comically pure. In their 1974 paper "Judgment under Uncertainty: Heuristics and Biases", Amos Tversky and Daniel Kahneman spun a wheel of fortune rigged to stop at either 10 or 65, then asked people to estimate the percentage of African countries in the United Nations. Those who saw 10 guessed a median of 25 percent. Those who saw 65 guessed 45. A number they had just watched being produced by luck, a number that could not possibly carry information, still dragged their estimates toward it.

You may ask: surely we only fall for that when we have nothing better to go on? Here is where it gets interesting. Nicholas Epley and Thomas Gilovich showed in their 2001 experiments that anchoring is not even one single mistake. When we generate the anchor ourselves, say starting from 1776 to work out when George Washington became president, we genuinely adjust away from it, but we stop adjusting too early, settling at the edge of what feels plausible. When someone else hands us the anchor, something different happens: our mind starts quietly building the case for that number, testing why it might be right rather than why it might be wrong. Two different mechanisms, one result. Which is why "just ignore the number" is not a strategy. The anchor does its work before you decide whether to take it seriously.

Anchoring is easy to confuse with confirmation bias, but they sit at different points in the timeline. Confirmation bias defends a belief you already hold, collecting friendly evidence after the verdict. Anchoring contaminates the estimate before you even notice you have formed one. By the time confirmation bias shows up to defend the number, the number was never yours.

So what can you do? I use a six-question reset. It will not delete the anchor, and I will not pretend otherwise. What it does is give the other numbers a chance to speak.

1. What is the anchor here? Name the first number, option, forecast, or confident opinion that entered this decision, and who put it there. An anchor you can name is an input. An anchor you cannot name is in charge.
2. What would I estimate from scratch? Before looking at the anchor again, build your own estimate from the ground up, and make it a range, not a point. This is first principles thinking applied to a number: strip the borrowed figure and ask what the components actually add up to.
3. What does the comparison class say? Find the base rate. What have similar projects cost, similar roles paid, similar plans taken in time? One relevant comparison class is worth ten rounds of adjusting someone else's figure.
4. What evidence would move me? Decide in advance what fact would push your estimate up or down. If nothing could move it, you are not estimating anymore, you are defending.
5. Am I comparing options on the same inputs? An anchor often survives by rigging the criteria. Put every option through the same decision matrix with independently generated numbers, and the first offer loses its home advantage.
6. What did I believe before the outcome? Write your estimate and your reasoning down in a decision journal before the result arrives. The anchor's favourite trick is convincing you, afterwards, that you knew it all along.

Run my three-thousand-dollar call through the reset and it collapses in minutes. The anchor was the client's opening line. My from-scratch range, built from hours and rate, started well above it. The comparison class, what similar projects had actually paid, agreed with my range and not with theirs. Nothing about the work had changed during that call. Only the frame had.

But hold on. Sometimes the first number deserves its influence. A market price condenses the judgment of thousands of buyers. An experienced colleague's estimate carries real information. The reset is not a machine for rejecting whatever arrives first. It exists to tell you which kind of number you are holding: one that carries evidence, or one that merely arrived early.

This is where ClearMind fits, quietly. It will not hand you the correct number, because there isn't one to hand. It works as a mirror: it asks where each figure came from, surfaces the assumptions your estimate is leaning on, and keeps your before-the-outcome reasoning where hindsight cannot redecorate it. The judgment stays yours. And since the anchor rarely whispers alone, the broader cognitive bias checklist is worth a pass before anything big.

In calligraphy, the first stroke decides where every later stroke can go. That is why the master pauses longest before it.

The first number in the room is not the truth. It is only the first guest to arrive.