Recency Bias: Is the Latest Event the Whole Story?

In one week last year, a long-standing client moved their brand work in-house, and two proposals I was sure about came back declined. By Friday evening I had half-written a note to my accountant about winding down the brand side of the studio. Then I opened the spreadsheet where I log every project. Over three years, brand work had been the steadiest line in the business. One bad week had not changed that. It had only become the loudest thing I could remember.

That is recency bias: giving the latest event more weight than it deserves, as if what happened last is what usually happens.

Memory research offers one possible clue. In 1962 the psychologist Bennet Murdock presented people with lists of words, one at a time, and asked them to recall as many as they could, in any order. The last several words came back best, the first few came back well, and the middle faded. A few years later Murray Glanzer and Anita Cunitz added a twist. If people spent half a minute on a counting task before recalling, the advantage of the final words disappeared, while the early words held up. Their reading was that the freshest items ride on a short-term store that fades quickly.

You may ask what word lists have to do with a business. Honestly, not much by themselves. They show something narrow: right after a list, the last items were the easiest to recall. Whether something similar colours how we weigh a bad week is a hypothesis, not a finding, but one worth testing on yourself.

Robin Hogarth and Hillel Einhorn studied something closer to real decisions: how people revise a belief as evidence arrives. In their 1992 model of order effects, order mattered, and how it mattered depended on the task. With short, simple sequences, judging after each new piece of evidence tended to favour what came last, while one judgment at the end tended to favour what came first. Longer or more complex sequences changed the picture. These were controlled scenarios, not a study of everyday life. Still, if you react after every update, it is worth asking whether the latest piece carries more weight than it has earned.

Recency bias is easy to confuse with its neighbours. Anchoring bias lets a first number or reference point pull every estimate after it. Recency is about weight, not reference points: the last event crowds the earlier ones out of view. Hindsight bias rewrites what seemed knowable once the outcome is in. Recency rewrites nothing. It turns up the volume on the newest event.

The fix is not to ignore new information, which is sometimes the most important thing you have. It is to put it back on the timeline. Here are six questions for when one recent event is pushing you toward a decision:

1. What exactly happened, and when? Write the latest event in one sentence with its date. Vague events grow. Dated ones stay their real size.
2. What time window does this decision live in? Choose the horizon before you look at the evidence. If the decision will run for three years, look at least three years back.
3. What does an equal window before it say? Compare the last month with the same month last year, or the months before.
4. What older evidence am I not remembering? List three things from early in the window, especially anything pointing the other way. That is first principles thinking in miniature: separating the latest observation from the whole period.
5. Would this event weigh the same if it had happened first? Imagine it at the start of the window, followed by everything since. If it shrinks, ask why. Sometimes newer evidence really is more informative. Sometimes it is only newer.
6. What would I need to see next to call it a pattern? Set the trigger in advance: two more quarters like this, or a second failed test. Then wait for the trigger, not the feeling.

Run my bad week through it. What happened: one client gone and two proposals declined, all in five days. The window: brand work was a three-year commitment, so three years. The equal window before: the same week a year earlier had brought in two new clients, which I had forgotten entirely. Older evidence: repeat clients made up most of the brand revenue, and the client who left had been hinting at an in-house team for months. Moved to the start of the window, the week read as a dip inside three good years. The trigger I set: two more quarters of falling brand revenue. They did not come. The note to my accountant went in the bin.

But hold on. Sometimes the latest event is the pattern. The six questions will not hide that. They only make sure you reach the conclusion with the whole window in front of you. If you are comparing options, score them in a decision matrix template using the full window. For a broader scan, the cognitive bias checklist covers the traps that travel with this one.

This is where ClearMind fits, quietly. It will not choose your time horizon or tell you whether the latest event matters. It works as a mirror: it guides you through questions about what you actually know, then reflects back the patterns in your answers and the blind spots you may have skipped. You choose the window. The decision stays with you.

There is an old Chinese story about a farmer whose horse ran away. His neighbours called it bad luck. He said, "Who knows?" The horse came back with another fine horse. Good luck, they said. "Who knows?" His son fell while riding it and broke his leg. Then the army took every able young man except his son. The farmer did not know the future. He simply refused to let the latest chapter stand in for the whole book.