Be decision-driven, not data-driven

- Be Decision-Driven Not Data-Driven — Techno Sapien - after eight years of measuring corporate data literacy, only 24% of companies report having reached data-driven nirvana. That’s fewer companies than last year. - decision-driven, not data-driven. The distinction may seem subtle, but it runs deep throughout a company’s culture. It’s also a false choice: decisions lead to action, but data informs action. - Decision-driven thinking differs from the data-driven approach in seven ways: - You start with questions, not data. Decision-driven teams spend more time designing questions and validating assumptions with decision-makers. Measure questions twice, cut data once! - Decision-makers lead projects, not data scientists. Like building a house, the homeowner (decision-maker) sets the tone, not the builder, architect, or plumber; they provide input. - You ponder the unknown more than knowns. Decision-driven thinking considers unknowns more than knowns. For example, in retail, a typical data-driven project optimizes existing loyalty programs; a decision-driven approach explores what makes customers happy in the first place, how attitudes are changing, and why. - You look wide first, then dive deep. Data-driven teams often dive head-first into the pool of data they already have. Puntonio and de Lange suggest you look “wide first, then narrow.” - You build new data boxes. Data-driven teams tend to think inside the box and invent ways to use the data they already have. When you start with decisions, teams more quickly discover missing or desired data. They spend time designing surveys, simulations, or searching outside the company for third-party data. - You’re more apt to spot and more able to reduce bias. By embracing a wider team upfront, decision-led teams tend to be more diverse. Diversity helps root out bias by questioning assumptions. - The business looks forward, not only in the rearview mirror. Data-driven thinking starts with historical data that accounts for what already happened. While the past might be prologue, pre-pandemic patterns might not still apply. - Pablo Picasso said, “the problem with computers is all they can do is provide answers.” His message is profound: don’t let tech lead; lead your tech. Be decision-driven, not data-driven.

Related: ![[datadriven.png]]

Notes mentioning this note

Data science is not scientific
Well, not all of it. Some of it is, but the unscientific bits are the most hyped, and the connections...
Data science is not scientific
Well, not all of it. Some of it is, but the unscientific bits are the most hyped, and the connections...

Here are all the notes in this garden, along with their links, visualized as a graph. If you don't see any nodes try zooming and panning in the grey area.

14 Points for ManagementAI is about making better, faster, cheaper...Abstractions raise tempoAct according to actual circumstancesAgile is the steering wheelAndon cords push decisions down the chainArchitecture sells optionsBayesian stats need simulationBayesianBe decision-driven, not data-drivenBe waterBeware of coordination headwindsBeware the ambiguity effectBlend multiple command systemsBuild and sellBullshit is more dangerous than lies to truthBusinessCarseCharacteristics of Special OpsChristopher AlexanderClimb the identity ladderClosed and open systemsComplexityConceptsContent actionability can be a crutchContradictions reveal answersData ScienceData science is not scientificDebt reduces optionalityDecision contexts can be complicated or complexDecisions matterDeleuzeDemingDo not tamperDon't forecastDon't look at the literature too soonDon't make single point forecastsDon't make type 3 errorsDruckerEinstein's DreamsEquality and freedom are opposedFacts dont change mindsFast languagesFire bullets, then cannonballsForecasting in human affairs is precariousFour laws of combatFoxes predict better than hedgehogsFunnel ExperimentGirardGo beyond ExcelHaikusHourly billing punishes expertiseHow Complex Systems FailHow to Get Startup Ideas, Jared FriedmanHow to Measure AnythingHow to Pitch Your Startup, Kevin HaleHow to Talk to Users, Eric MigikovskyHow to do philosophyHuainanziIOHAIIdentifying the Centre of GravityIllusion of objectivityIn complex contexts, simulateIn complex contexts, use AI as sensorsIncentives matterIncrease your tempoInman's rulesIntelligence Analysis FrameworkJockoJohn BoydKeep it simpleKeep movingKeep your identity smallKnow thyselfLeadershipLevel productionLeverage PointsLibertarianismLinear roadmaps are a lieLittle big foreign phrasesLittle big phrasesMBO ignores processesManeuver warfareManufactured fun is no fun at allManufactured fun is not funMarketing AttributionMarketing is designMessy thought, neat thoughtMilitaryModel Questions vs Actor QuestionsMost things aren't treesMuddle throughName three alternativesNations don't trade with each other, people doNever do great thingsNobody knows anythingNon-systemic thinking is LAMONothing in excessOKRs aid EinheitOKRs are not that greatOODA LoopsOmit needless wordsOperating on the front is the way to use...Oren KlaffPaul Graham et alPeople are not the problemPersuasionPhilip MorganPhilosophyPointersPoints of viewsPolicyPournelle's iron law of bureaucracyProbes over experimentsProductivityProducts are functionsQuestions revealQuotesReality has a suprising amount of detailRecognizing the two times of schedules is central...Repeated games are different from single turn onesRithwikRyan Singer on Christopher AlexanderScreen sharing is the modern GembaSocial networks are like ICOsSolutionismSometimes daily billing makes senseSpeed mattersStart with actionState of Play, 20 questions, VGRStoicismStory Structure 101StrategyStrong opinions, weakly heldSubtractSun TzuSystems over goalsSystemsTeam of TeamsThe five elements of an organization that can...The three languages of politicsThe use of AI can be strategicThere are trade-offs everywhereThere is a robustness-efficiency tradeoffThere will be a last timeThink in publicTiagoTimeToyota Production SystemTrainability wins at scaleTraining helps in calibrations and predictionsUse more Saxon words to be clearVGRVia negativaWanting has layersWhat is Decision EngineeringWhen AI met StrategyWittgenstein's RulerWords matterWrite code, talk to usersWriting heightens consciousnessWritingYCombinator Startup SchoolYou need less dataYou need more data14 Points for ManagementAI is about making better, faster, cheaper...Abstractions raise tempoAct according to actual circumstancesAgile is the steering wheelAnalysis and SynthesisAndon cords push decisions down the chainArchitecture sells optionsBayesian stats need simulationBayesianBe a Greek, not a BabylonianBe decision-driven, not data-drivenBe waterBeware of coordination headwindsBeware the ambiguity effectBlend multiple command systemsBuild and sellBullshit is more dangerous than lies to truthBusinessCRAP framework for dealing with workplace bullshitCarseCausalityCharacteristics of Special OpsChristopher AlexanderClimb the identity ladderClosed and open systemsComplexityConceptsContent actionability can be a crutchContradictions reveal answersData ScienceData science is not scientificDebt reduces optionalityDecision contexts can be complicated or complexDecisions matterDeleuzeDemingDesignDo not tamperDon't forecastDon't look at the literature too soonDon't make single point forecastsDon't make type 3 errorsDruckerEinstein's DreamsEquality and freedom are opposedFacts dont change mindsFast languagesFire bullets, then cannonballsForecasting in human affairs is precariousFour laws of combatFoxes predict better than hedgehogsFunnel ExperimentGirardGo beyond ExcelHaikusHourly billing punishes expertiseHow Complex Systems FailHow to Get Startup Ideas, Jared FriedmanHow to Measure AnythingHow to Pitch Your Startup, Kevin HaleHow to Talk to Users, Eric MigikovskyHow to do philosophyHuainanziIOHAIIdentifying the Centre of GravityIllusion of objectivityIn complex contexts, simulateIn complex contexts, use AI as sensorsIncentives matterIncrease your tempoInman's rulesIntelligence Analysis FrameworkJTBD by data teamsJockoJohn BoydKeep it simpleKeep movingKeep your identity smallKnow thyselfLeadershipLevel productionLeverage PointsLibertarianismLinear roadmaps are a lieLittle big foreign phrasesLittle big phrasesLoreMBO ignores processesManeuver warfareManufactured fun is no fun at allManufactured fun is not funMarketing AttributionMarketing is designMessy thought, neat thoughtMilitaryModel Questions vs Actor QuestionsMost things aren't treesMuddle throughName three alternativesNations don't trade with each other, people doNever do great thingsNobody knows anythingNon-systemic thinking is LAMONothing in excessOKRs aid EinheitOKRs are not that greatOODA LoopsOmit needless wordsOperating on the front is the way to use...Oren KlaffPaul Graham et alPeople are not the problemPersuasionPhilip MorganPhilosophyPointersPoints of viewsPolicyPournelle's iron law of bureaucracyProbes over experimentsProductivityProducts are functionsQuestions revealQuotesReality has a suprising amount of detailRecognizing the two times of schedules is central...Repeated games are different from single turn onesRethink design thinkingRithwikRyan Singer on Christopher AlexanderSailing over rowingScreen sharing is the modern GembaSocial networks are like ICOsSolutionismSometimes daily billing makes senseSpeed mattersStart with actionState of Play, 20 questions, VGRStoicismStory Structure 101StrategyStrong opinions, weakly heldSubtractSun TzuSystems over goalsSystemsTeam of TeamsThe Expert's TrilemmaThe five elements of an organization that can...The three languages of politicsThe use of AI can be strategicThere are trade-offs everywhereThere is a robustness-efficiency tradeoffThere will be a last timeThink in publicTiagoTimeTo think is to forgetToyota Production SystemTrainability wins at scaleTraining helps in calibrations and predictionsUse more Saxon words to be clearVGRVia negativaWanting has layersWhat is Decision EngineeringWhen AI met StrategyWittgenstein's RulerWords matterWrite code, talk to usersWriting heightens consciousnessWritingYCombinator Startup SchoolYou need less dataYou need more data