Collecting

Open-ended learning, before anyone has a hunchEvery design decision needs something to stand on. Collecting is where that foundation gets built: the team gathers evidence about its audience broadly and openly, from the sources it can reach, before anyone has decided what they believe.User Needs names what users are trying to do. Audience defines who you learn from and how much each source can be trusted. Collecting reaches those people and gathers what they do, say, and feel, documented well enough that anyone downstream can trust where it came from. Situations then finds the recurring conditions in what was gathered.Collecting is not testing. Testing, in Measure, starts because the team has a hunch about a concept. Collecting comes first and stays open. An interview with no hunch behind it is Collecting; an interview probing a specific hunch is Testing. An analytics feed is Collecting; reading that feed against a hunch is Testing.What This Block CoversCollecting is organized into seven sections, each with a specific job:SectionWhat It DoesOverviewExplains what Collecting is, where it sits in Define, and how it differs from Testing in Measure.TechniquesCovers the open-ended gathering techniques and the ongoing monitors that keep evidence flowing.PlaybookProvides the four-step collection process, inputs, prompts, and the Source Map.ReferencesProvides the Sources catalog, the role-tagged Tools list, evidence from the Glare MCP, and key terms.ExamplesShows collection situations, near misses, and common traps.DecisionsHelps teams identify their situation and choose the next step.Agent OperationsDefines how AI Skills behave in Collecting: routing, confidence, and output contracts.What Collecting SolvesMost teams form beliefs from whatever evidence is closest: one loud customer, last quarter's survey, the dashboard someone happened to open. Their hunches then rest on a narrow base, and the tests that follow measure the wrong things well.Collecting widens the base on purpose. It reaches the defined audience through sources that can represent them, gathers more than one kind of evidence, and records where each piece came from. The result is a map of what the team knows, how far it can trust each source, and where the gaps are.Evidence nobody can trace is worse than none, because it produces confidence without a foundation. Recorded sources are what let Situations read the evidence later without guessing where it came from.What Kind of Evidence Are You Gathering?Gather more than one kind, so the picture includes what people do and what they say.Data TypeWhat It CapturesExamplesView user dataBehavior patterns and usage over timeAnalytics, clickstream, funnels, heatmapsSee what users doHow people work in their own contextA day in the life, diary studies, session recordingsSense what users feelEmotional response and satisfactionOpen satisfaction surveys, reviews, sentiment in support ticketsHear what users sayOpinions, expectations, and languageInterviews, needfinding surveys, forums and community threadsPair at least two types. Behavioral evidence shows what happened; attitudinal evidence shows what people report. Neither is interpreted here. Both are recorded so Situations can read them together.The Four StepsEvery collection effort in Glare follows the same four steps:Name what you want to learn.Keep the learning open, but aim it at a user need or an audience. If a belief is already there, write it as a hunch instead.Choose sources.Pick sources that can reach the defined audience, note how credible each one is, and check what evidence already exists.Gather.Collect openly with more than one kind of evidence, and tie each piece to its source. Check which monitors are already running and what the Glare MCP can already return.Map what you know.Record what was gathered, how far to trust each source, and where the gaps are.The output is a Source Map: the learning goal, the need or audience it is about, each source with its credibility, the techniques and data types used, the evidence, and the gaps. It holds no hunches or signals. Collecting hands it to Situations, which finds the recurring conditions in it.The Glare MCP and CollectingThe Glare MCP is the main way to widen Collecting beyond what a team could read by hand. It returns user needs, situations, audience detail, and expert perspective drawn from transcripts, forums, reviews, articles, and studies, each with its source attached. In Measure and Focus, the same evidence becomes a data source that hunches are tested against and options are compared on.The Line Between Collecting and TestingThe line is the hunch. Collecting comes first and stays open. Testing, in Measure, starts once the team believes something specific about a concept. No hunch: Collecting. A hunch: Testing.WorkNo hunch yet: CollectingA hunch: TestingInterviewsOpen interviews about goals, work, and frustrationsInterviews probing a specific hunchAnalyticsAn ongoing monitor, such as a funnel reviewReading that feed against a hunchOutside evidenceGlare MCP evidence gathered to learn what existsThe same evidence used as a data source to test a hunchCollecting keeps two kinds of technique: open-ended gathering, such as interviews, a day in the life, diary studies, needfinding surveys, listening, and support tagging, and ongoing monitors, such as funnel review and user metrics. Techniques that test a design or a hunch, such as first click, task success, card sorting, and SUS, live in Testing. A/B and multivariate tests compare options, so they live in Focus. The full list is in Techniques.Proof in PracticeA university team was stuck on a navigation decision. Meetings circled for weeks. Everyone had an opinion. Nothing moved.The debate ended when the team ran a preference test through Helio on two navigation options. That was Testing in Measure, not Collecting: the team already had two options and a belief about which would work.Collecting is what should come before a debate like that. Open learning about how the audience looks for things, from sources that reach them, gives the options a base in evidence instead of opinion. The hunch that follows then has something real to stand on.Business ImpactA wider evidence base means fewer hunches built on one anecdote.Recorded sources let anyone trace a later signal back to where it started.Ongoing monitors catch changes early, before they reach a quarterly review.Evidence from the Glare MCP makes open-ended learning cheaper, so teams do more of it.Collecting is the foundation of measurable design confidence. Every dataset captured well is a small act of momentum.Where Collecting Sits in DefineCollecting is the third block in the Define area, sitting between Audience and Situations. User Needs and Audience tell you what to learn and who to learn it from. Collecting reaches those people, gathers openly, and hands a Source Map to Situations. Situations finds the recurring conditions in what was gathered and reads them as signals.BlockWhat It Focuses OnUser NeedsThe motivations, expectations, and goals that drive behavior.AudienceThe people you learn from, and how much to trust each source.CollectingReaching the audience and gathering evidence openly, with sources recorded.SituationsFinding the recurring conditions in what was gathered and reading them as signals.When a belief forms out of what was gathered, it leaves Define. It becomes a hunch about a concept in Measure, and Testing checks it.AI PromptThis prompt helps you plan open-ended learning about a user need or an audience, before anyone has a hunch.Start with what you want to learn and who it is about. It guides you to:Name an open learning goal tied to a user need or an audienceChoose sources that can reach the audience, and note how far to trust each onePick open-ended techniques and monitors that cover at least two kinds of evidenceMap what you know, how far to trust it, and where the gaps areYou'll end with a Source Map that's ready to hand to Situations, with sources, techniques, evidence, and gaps all named.Use this before any fieldwork begins, and before anyone writes a hunch.AI SkillsThe Collecting skill file teaches your AI the four-step collection process so it can plan open-ended learning for any need or audience.Load it when you need to go deeper on choosing sources, pairing kinds of evidence, or building a Source Map. It gives your AI:The four steps, from naming what you want to learn to mapping what you knowThe open-ended gathering techniques and ongoing monitors, with the data type each coversThe Source Map fields and the handoff to SituationsThe rule that keeps Collecting open: no hunch here, and a hunch goes to Testing in MeasureDownload the skill file below to use the full Collecting framework with your AI assistant.