Somewhere in your last month of work, a number was said out loud, and you have been quietly obeying it ever since. Maybe a manager asked, "Could this be done by Friday?" Maybe a recruiter mentioned a salary range before you had said a word. Maybe you glanced at last quarter's figure while drafting this quarter's plan. You didn't decide to follow it. You simply never noticed that the number had already done most of your thinking for you.

This is anchoring, and it is one of the most durable findings in behavioral science. It is also one of the least comfortable, because it works even when you know about it, and even when the number is absurd.

What the research actually shows

In the early 1970s, Amos Tversky and Daniel Kahneman asked people to estimate the percentage of African countries in the United Nations. Before answering, each person watched a wheel of fortune land on a number. The wheel was rigged to stop on either 10 or 65. People who saw 10 gave noticeably lower estimates than people who saw 65. Everyone knew the wheel was random. It moved their answers anyway.

That is the core of the effect: an initial value, even an irrelevant one, pulls later judgments toward itself. Tversky and Kahneman proposed that people start from the anchor and adjust, and that the adjustment is usually insufficient. You stop moving as soon as the answer feels plausible, not when it is right.

A second explanation, developed by Thomas Mussweiler and Fritz Strack, is selective accessibility. When you consider whether an answer might be near the anchor, your mind goes looking for evidence that fits it, and that evidence then feels like your own reasoning. The anchor doesn't just nudge the number. It quietly edits which facts you remember.

And it reaches experts. In a well-known study by Gregory Northcraft and Margaret Neale, real estate agents toured a house and were given a packet that included a listing price. Agents handed a higher listing price appraised the house higher than those handed a lower one. Most denied that the listing price had influenced them.

Where it hides in an ordinary workday

Once you know the shape of it, you start seeing anchors everywhere.

Estimates. Someone says "this should take a couple of hours," and your own estimate orbits that figure, even if you have never done anything like it. The first guess in the room is rarely the most informed one, but it becomes the center of gravity.

Negotiation. Research on negotiation, including work by Adam Galinsky and Thomas Mussweiler, found that the party who makes the first offer tends to end up with a result closer to their opening figure. Silence feels polite. It often hands the anchor to the other side.

Your own past. Last year's budget, last sprint's velocity, the hours you spent on the previous report. Self-generated anchors are still anchors. They feel like data, but they are often just the first thing your mind grabbed.

Your self-worth. The quieter version is personal. A single harsh remark in a review, or one number on a scale or a pay stub, becomes the reference point against which everything else is measured. Praise gets discounted as adjustment. The criticism stays the anchor.

Why knowing about it doesn't cure it

The frustrating part is that awareness alone is a weak defense. Studies of debiasing find that anchoring is stubborn precisely because it operates before deliberate thought. You can't simply decide to ignore a number you've already read.

What does help is changing the process rather than trusting your resolve. The most reliable approach in the literature is consider the opposite: deliberately generating reasons why the anchor might be wrong. Mussweiler and colleagues found that asking people to think about why an anchor might be inappropriate reduced its pull. You are giving your memory a second set of evidence to search, so it doesn't just gather the first set.

The other approach is to get a competing reference point. A single anchor dominates. Two or three independent ones give you a range to reason within.

Your next moves

  • Estimate before you listen. Next time you're asked how long something will take, write your own number privately before anyone speaks, or ask for a few minutes before answering. Even a rough private figure is a defense against the first voice in the room.
  • Ask for the opposite case. Before accepting any figure, write one sentence: "This number could be too high because..." and one for too low. Do both. It takes two minutes and it is the best-supported fix we have.
  • Collect three references. For any estimate or price, find three unrelated comparisons: a similar past task, someone who has done it, a market rate. Use the range, not the first value.
  • Make the first offer when it's your move. In a salary or scope conversation, do your research ahead of time and open with a specific, justified figure rather than waiting. Be prepared with a reason, since a number with a story attached holds better.
  • Audit one inherited number today. Pick a deadline, budget, or target you've been carrying forward and ask, "Who set this, and what was it based on?" If the honest answer is "nobody, it was just first," rebuild it from scratch.

One more thing about the numbers in your head

Much of what feels like your own judgment is really a conversation with whatever was said first. Zenith is built around the idea that a day's plan should come from what you actually have in front of you, your tasks, your energy and your real time, rather than from the loudest figure in the room. If you'd like a calmer place to build estimates on purpose, you can try it at https://zenith.lumenlabs.works.