Cognitive Biases · CB-50
People search for answers where it's easiest to look, not where the answer is actually most likely to be found — like searching for lost keys only under the streetlight, because that's where you can see.
A bias in how people (and organizations) gather evidence, in which they preferentially search for information in places that are easy to observe or measure, rather than in the places most likely to actually contain the relevant answer — named for the old joke about a man searching for his lost keys under a streetlight, not because he dropped them there, but because that's where the light is.
A long-circulating folk parable formalized as a named cognitive/methodological bias in decision science and research-methods literature, frequently invoked to describe a systematic mismatch between where evidence is sought and where it's actually needed.
The Mechanism
Searching where it's easy to see, not where the keys actually are
The parable's man never finds his keys, no matter how thoroughly he searches under the streetlight — because the ease of searching a well-lit area has no bearing on whether the answer is actually located there, a mismatch that recurs constantly in research, business analytics, and public policy whenever available data is confused with relevant data.
01 · IT'S DRIVEN BY THE AVAILABILITY OF DATA, NOT THE RELEVANCE OF DATA
Easy-to-measure proxies get substituted for hard-to-measure true answers
Organizations and researchers frequently default to analyzing whatever data is already collected and readily available, rather than the data that would actually answer the question at hand — the well-lit, measurable proxy becomes the de facto focus of analysis, regardless of how well it actually represents the underlying question.
02 · IT PRODUCES CONFIDENT-LOOKING ANALYSIS THAT ANSWERS THE WRONG QUESTION
Thoroughness in the wrong place can look more rigorous than honest uncertainty about the right question
A large, well-lit search under the streetlight can produce an impressively thorough-looking body of analysis, creating a false sense of rigor and confidence — even though the actual question of interest remains entirely unaddressed, because the truly relevant evidence was never in the searched location to begin with.
03 · IT'S A RECURRING PROBLEM IN MEDICAL, SOCIAL SCIENCE, AND BUSINESS RESEARCH WHERE THE BEST DATA ISN'T THE MOST RELEVANT DATA
Research and business analytics gravitate toward whatever's already digitized and quantifiable
Medical research has been criticized for over-relying on easily quantifiable biomarkers over harder-to-measure but more clinically relevant patient outcomes; business analytics teams often default to whatever metrics are already instrumented in existing systems, rather than the harder-to-capture metrics that would actually answer the strategic question at hand.
Where It Fails / Inversion
Where it fails / inversion
Sometimes the easily available data genuinely is the most relevant data, and dismissing convenient, well-measured evidence purely because it was convenient to obtain is itself an error — the actual test is whether the available data genuinely answers the question, not whether it happened to be easy to gather; ease of measurement and relevance aren't mutually exclusive, just not reliably correlated.
How To Use It
Worked example · diagnosing why a product isn't selling
A team analyzing why a product underperforms, using only the extensive web analytics already instrumented on their site (which pages get visited, click-through rates), may be searching entirely under the streetlight if the actual reason customers aren't buying lies in offline conversations, competitor comparisons, or word-of-mouth concerns that were never being measured at all — a deliberate effort to gather the harder-to-collect but more relevant evidence (direct customer interviews, lost-deal analysis) is often needed to find the real answer.
How to use it
Before trusting an analysis based on the most readily available data, explicitly ask whether that data is actually the most relevant evidence for the question at hand, or simply the easiest to measure. If the truly relevant evidence would require more effort to gather, that effort is often exactly what's needed rather than optional.
See Also