Recency bias runs on how memory works under time pressure. The most recent piece of evidence is the easiest one to recall, so it feels like the most relevant one, even when it isn't. Judgement follows memory more closely than most leaders assume.
For a leader, this shows up as a decision that tracks whoever spoke last, or whatever happened most recently, rather than the fuller pattern. A quarter of steady results can be overturned by one bad week, and a year of mixed performance can be forgiven because of one strong final month. The record didn't change; what you noticed last did.
Recency bias is a close relative of outcome bias, though the two distort judgement from different angles. Outcome bias judges a decision by how it turned out. Recency bias judges a whole period by whatever turned out last, regardless of how representative that moment actually was.
The practical tell is timing. If a judgement was reached right after a strong or weak result landed, and would have been different a fortnight earlier or later, the most recent event was doing the deciding, not the fuller record.
Our monthly review meetings with Tom ran on whatever had happened the week before. A good week meant a good meeting: numbers, energy, no real scrutiny. A poor week meant the opposite, no matter how the month as a whole was actually shaping up. I'd bring the trend line every time, hoping to shift the conversation onto the fuller picture, but the previous week's numbers had already set his mood and his judgement before I'd said a word.