Naming what changes between measurementsVariables name the one thing that changes between measurements. Measure runs one test to get an answer. Focus repeats that same measurement, with the same UX metric, technique, and audience, while one thing changes: the version, the competitor, the segment, the time period. The variable is that one thing.Choosing it is like choosing an audience in Define or forming a hunch in Measure. It is the setup the team picks before it acts. Once the variable is named, Comparing repeats the measurement across it, and Bets turn the difference into a bet the team can act on.SectionPurposeOverviewExplains what a variable is and its role inside FocusTechniquesCovers how to find the variable that fits a decision, including the mapping methods that surface itPlaybookProvides operational workflows, inputs, prompts, outputs, and execution guidanceReferenceContains the 12 variables, what changes in each, and the questions each one answersExamplesProvides worked examples and a pattern library for practitioners and AI-assisted workflowsFormatsPrepares a variable choice for a specific purpose, audience, and formatDecisionsHelps teams recognize when Variables is the right place to work and where to startAgent OperationsDefines how AI Skills consume the block, route work, and preserve judgmentOne variable at a timeA comparison is only fair when one thing changes. If the version and the audience both change between two tests, the team cannot say which one caused the difference. Naming the variable keeps everything else fixed, so the result can be trusted and explained.A strong variable:Is the one thing that differs between measurementsConnects to the decision the team needs to makeKeeps the UX metric, technique, and audience the same on every sideCan be repeated, so the next round lines up with the lastThe 12 variablesEach variable names a different kind of change. Pick the one that matches the decision in front of the team.VariableWhat changesQuestion it answersIterationVersions of the same designDid the new version improve the signal?CompetitorsYour experience against othersWhere do competitors create more clarity?TimelineThe same measure across time periods: baseline, launch, +90 daysAre signals improving, and did the change hold?SegmentsAudiences, roles, or cohortsWhich group sees the most value, and which struggles?JourneysSteps or touchpoints, with the same metric at each stepWhich moment creates the most friction?Platforms / DevicesMobile, desktop, tablet, or channelDoes the experience hold up everywhere?User Goals / TasksDifferent tasks on the same experienceWhich task is easiest or hardest?Feature UsageFeatures compared on value and useWhich features create value, and which are ignored?GeographiesRegions, markets, or languagesWhere does it perform better or worse?User LifecycleNew, returning, and expanding usersWhat does a new user need compared with an existing one?Behavioral TriggersPrompts or conditions that cause actionWhat makes users start, continue, or hesitate?SeasonTiming or cycles that change needsIs the signal durable, or tied to timing?Finding the variableSometimes the variable is obvious: two versions, two segments, before and after a launch. When it isn’t, a mapping method can surface it. Mapping methods shape the work so the team can see what is worth comparing. They don’t measure anything themselves.Journey mapping and service blueprints:show which steps or touchpoints to compare.Empathy mapping and Jobs-to-be-Done:show which segments or goals differ.Lifecycle stage mapping:shows where new and existing users split.Trigger and habit loop mapping:show which conditions to vary.Opportunity Solution Tree:shows which options are worth versioning.Where the other methods wentMethods used to name this block and mixed several kinds of tool. Each kind now lives where it does its work:KindWhat it doesWhere it lives nowMeasuringCollects one data point you can repeat: task success, first click, SUS, an A/B result used as a measureMeasure → Testing, or Define → Collecting for ongoing monitorsComparison designsSet up the repeat: randomized A/B, before and after, benchmark, segment splitFocus → Comparing → TechniquesMappingShows what to compare: journey mapping, JTBD, empathy mapping, service blueprintVariables, as ways to find the variablePrioritizingRanks options after the evidence is in: RICE, ICE, LNO, MoSCoW, Eisenhower, Now/Next/LaterFocus → BetsBet sizingSizes the commitment to the evidence: Four Big Risks, smoke test, pilotFocus → BetsWhere Variables fit in GlareVariables sit inside the Focus area of Glare, between Initiatives and Comparing. Initiatives define the area of work that needs attention. Variables name what changes. Comparing repeats the measurement across the variable. Bets turn the evidence into action.StageRoleInitiativesDefine what needs attention and what the work is trying to improveVariablesName the one thing that changes between measurementsComparingRepeat the same measurement across the variable and read the differenceBetsRank the options and commit to a directionVariables are how Focus keeps comparisons fair. They name what changes, so everything else can stay the same.AI PromptThis prompt helps you name the one thing that should change between measurements, rather than defaulting to the nearest test.Start with a named initiative and the decision your team needs to make next. It guides you to:Ask what you would need to see change to make the callLook at the data already available along that variable before collecting morePick one of the 12 variables and the options along itList what stays fixed: the UX metric, technique, and audienceYou'll end with a named variable, the options to compare, and what stays fixed, ready to hand to Comparing.Use this after an initiative is scoped and before any comparison is set up, when the team needs to agree on what changes.AI SkillsThe Variables skill file teaches your AI the 12 variables and the one-variable rule, so it can help you pick the comparison that fits the decision rather than the test most familiar to the team. It still runs as glare-focus-methods, since this block used to be called Methods.Load it when you need to find the variable for a decision, sort a list of methods into where they belong now, or check a planned comparison for more than one change. It gives your AI:The 12 variables, what changes in each, and the question each one answersThe one-variable rule, and how to split a comparison when two things changeThe mapping methods that help find the variable when it isn't obviousWhere the old methods went: measuring, comparison designs, prioritizing, and bet sizingDownload the skill file below to use the full Variables block with your AI assistant.
Variables
Related links
LogRocket guide explaining what a prototype is and walking through feasibility, low-fidelity, high-fidelity, and data-driven prototypes. Useful when picking the right prototype type for a question you want to answer.
Argues the MVP is the wrong tool for finding weak spots and offers RAT — the Riskiest Assumption Test — which scores each assumption by probability of being wrong times impact, then tests the top one. Useful when a team is racing to build an MVP but has not asked which single assumption could kill the idea.
Cheat sheet comparing empathy maps, journey maps, experience maps, and service blueprints, with notes on when each is best. Useful when picking a map type for a project and you need a quick side-by-side reference.
Identify where decision quality breaks down
The Glare Design Assessment helps teams spot weak validation, stakeholder friction, alignment gaps, and assumptions that scale without measurable learning—so you have a clearer starting point for improvement.
About 5 minutes · Team-based · Diagnostic snapshot you can act on
Take the Design Assessment
