01Big decisions, tiny samples
A career switch, a move across the country, leaving a job to go independent — these are among the highest-stakes choices people make, and they’re usually decided on a handful of anecdotes from whoever happens to be in reach. A sample size of three, weighted by who told the most vivid story.
It’s not that people don’t want more evidence; it’s that gathering it feels impossible. But for almost any major life decision, hundreds of people have already made the same call and written candidly about how it turned out.
02Someone already lived your decision
Communities exist for nearly every fork: people who went freelance, relocated to the city you’re eyeing, took the certification, had the surgery. They describe what they expected, what actually happened, and whether they’d do it again — unprompted, and with the detail people only share when no one’s selling them anything.
That’s a body of lived experience far larger and more honest than your immediate circle can offer.
03What to look for
Reading others’ experiences to inform your own, hunt for:
- People who did it and regret it — and the specific reason why
- People who did it and wouldn’t change a thing — and what made it work
- The surprises — the things nobody warned them about beforehand
- The conditions — what was true for those it worked out for
- The timeline — how they felt right after vs a year on, since first reactions mislead
04Reading lived experience at scale — with care
The pipeline can pull the threads where people recount the decision and score the sentiment and intensity across them, turning scattered stories into a readable pattern: how often regret shows up, what conditions separate the happy from the disappointed.
Hold two caveats firmly. Online accounts skew toward the strongly satisfied and the strongly regretful — the quietly content rarely post — and no aggregate can tell you about your specific circumstances. Use it to widen your evidence well beyond the three people you’d otherwise ask, not to outsource a decision only you can weigh.
Check whether it’s really widespread
The same trap — deciding from too small a sample — applies to any problem you feel strongly about.
Is it just me? Checking if a problem is widespread →05A worked example: pressure-testing a market-entry bet
This walk-through is illustrative — the point is the method, not the specifics. Say you run a small project-management tool and you are weighing a big move: building a dedicated mobile app because you believe your users are clamouring for one. Here is how you would stress-test that before committing months of build time.
- State it as a falsifiable claim. Not “a mobile app would be nice” but “our users actively want a native mobile app and will use it for real work.” A vague hope cannot be tested; a sharp claim can.
- Hunt for the disconfirming evidence first. Before collecting cheerleaders, go looking for the people who built a mobile version and regretted it, or who tried competitors’ apps and abandoned them. If the claim survives a deliberate search for reasons it is wrong, it is far stronger.
- Count independent voices, not loud ones. One detailed thread with forty people agreeing beats a single vivid demand from your most vocal customer. Weigh how many separate, unconnected people support the bet versus how many contradict it — and discount duplicates and obvious self-promotion.
- Separate “would be nice” from “would actually use.” People say they want apps far more readily than they open them. Look for accounts describing real, recurring mobile use of similar tools, not stated enthusiasm.
- Decide with eyes open. If support outweighs contradiction across many independent voices and the disconfirming cases do not apply to you, proceed — knowing the specific risks you found. If the loudest demand turns out to be three people, you have just saved a quarter of engineering time.
Frequently asked questions
Turn the decision into a sharp, falsifiable claim, then go looking for evidence that it is wrong before you collect support. Weigh how many independent, unconnected people back it versus contradict it — discounting loud or self-interested voices — and check that stated enthusiasm matches real behaviour. Proceed only when support clearly outweighs the disconfirming cases.
The cheapest risk reduction is reading what happened to people who already made the same call. Search deliberately for the regret stories and the surprises nobody warned them about, not just the success cases. The conditions that separated the happy from the disappointed tell you whether your own circumstances line up — long before you spend money or time committing.
Write the assumption as a claim someone could prove false, then try hard to falsify it. If you assume customers want a feature, search for people who got that feature elsewhere and ignored it. An assumption that survives a genuine attempt to break it is worth acting on; one that collapses on first inspection just saved you the build.
You want a body of independent accounts large enough to see a pattern, not three anecdotes from people you happen to know. Look for both sides — the satisfied and the regretful — the surprises they did not expect, and the conditions that made it work. Crucially, weigh revealed behaviour over stated intent, since people say more than they do.