Sample frameDay 066·Decisions · Bias

Good morning. The example that arrives first in your head is the best-told story you know, not the most likely thing to happen next.

Decisions.

Availability Bias

5 min read·Apply by lunch

The question

Think about the risk your team has spent the most energy guarding against this quarter: is it genuinely the most likely thing to damage you, or simply the most recent thing to damage somebody you know? If the answer will not come without going and counting, you already have it — the quarter has been spent defending against a story you can retell rather than a number you have checked.

The idea

In 1978 the psychologists Sarah Lichtenstein, Paul Slovic, Baruch Fischhoff and colleagues, working in Oregon, published 'Judged Frequency of Lethal Events' in the Journal of Experimental Psychology: Human Learning and Memory. They asked people to estimate how many Americans die each year from around forty different causes, then set the answers against the actual figures. The errors were not random. Dramatic, heavily reported deaths — tornadoes, floods, homicide, botulism — were wildly overestimated, while the quiet common killers such as stroke, asthma and diabetes were underestimated. The estimates tracked press coverage rather than mortality. That is availability, the heuristic Amos Tversky and Daniel Kahneman had named five years earlier: the mind answers 'how likely is this?' by testing how easily an example comes to mind, and ease of recall runs on vividness, recency and repetition, none of which is frequency. Your business version is the deal you lost last month, the hire who blew up, the competitor who just raised. Recency is not probability, and memory is not a dataset.

The framework

01
Catch the estimate as it forms — then ask what it is made of.
Most availability errors enter the room as a confident sentence: 'customers always ask for this', 'that channel never works for us', 'enterprise deals take a year'. Every one is a frequency claim wearing the clothes of experience. When you hear one, including from yourself, stop and ask what the estimate is built from: how many cases, over what period, and how they came to be the ones you remember. The honest answer is usually two or three, all recent, all memorable, at least one still slightly painful. That is not a sample, it is a highlight reel, and the room is about to build a quarter on it.
02
Go to the record before you go with the memory.
The correction is unglamorous and it is the entire discipline: count. Before anyone argues about how often something happens, spend ten minutes inside a record you already own — the CRM, the invoice history, the support log, last year's calendar — and produce an actual number. Memory retrieves by salience; a record retrieves by date, which is the only sampling method that can answer a frequency question. A thin record still beats recollection. And where no record exists, refuse the anecdote as a substitute: write down the number your decision needs to be true, then go and find out whether it is.
03
Discount every example for how loudly it was told.
Weight each case by the route it took into your head. Recency, personal involvement, financial pain, public embarrassment and sheer retelling all make an example easier to recall while making it no more likely to recur. The effect runs deeper than it sounds. In 1991 Norbert Schwarz and colleagues found that people asked to recall twelve examples of their own assertive behaviour rated themselves less assertive than people asked for only six: the difficulty of the recall, not the number recalled, became the evidence. Fluency is information about your memory, not about the world. When an example arrives instantly, treat the speed as a warning rather than a confirmation.
04
Fix the input, not just the conclusion.
What comes to mind is whatever you have recently fed yourself, so manage the diet rather than arguing with the output. Most operators run on an input mix built almost entirely from complaint: the escalation inbox, the loudest customer, the competitor announcement, the deal that collapsed on Friday. None of it is false and all of it is unrepresentative. Build the counterweight on purpose — read a sample of won deals beside the lost ones, ring two customers who left without saying anything, and keep a standing register of the outcomes you keep having to estimate, so next quarter's judgement is a lookup rather than a retrieval.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

Availability Bias is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Decisions frameworks

All 28 decisions frameworks →