Bets

Bets turn compared options into a commitment.By the time a team reaches Bets, the work should no longer be floating in opinion. The initiative is clear. The variable is named. Comparing has repeated the same measurement across the options and shown which one works better, and which one people would choose.Now the team commits. A bet names the option it backs, the evidence behind it, what it costs, and the outcome it expects. Without a bet, comparisons stay in decks and the same options come back to every meeting.What Bets answersEach facet ends in a block that judges one quality, as people experience it. Bets closes Focus, and the quality it judges isdesirability: "this is the outcome I'd choose."Question it answersWhat it checksWhat it hands upOf the options, which outcome is worth committing to?Did the chosen option win on its outcome measure, and at what cost?A committed bet with a predicted outcomeBets is accountable forPrefersignals: whether people choose one option over the others. Comparing usually produces Use signals too, such as which option people could find or finish. A bet needs both: Use says the option works, Prefer says people want it.Desirability here is the quality the block judges. It is not the same thing as the Desirability UX metric. That metric can feed a Prefer signal, but on its own it doesn't show a choice between options.Why Bets MatterMost teams don't struggle for ideas. They struggle to commit. One stakeholder wants to ship, another wants another round, and the meeting ends where it started.A bet settles it. It helps teams:choose between options on evidence, not on who argues longestsee whether people actually prefer the option the team likesname the cost of the choicewrite down the outcome the business should seehand Lead something it can check laterEvery comparison carries a Prefer signalA bet is only as good as the evidence that people would choose it. Every comparison readout that leads to a bet follows one rule.1.Measure Prefer the same way for every option.Use one of these: choice between options shown side by side, likely to choose (top-two box), trust that it will turn out as shown, or pull-away (would leave or look elsewhere). Desirability words count only next to one of these.2.Give it its own signal, tagged Prefer,with every option's number side by side.3.State a Use line and a Prefer linein the bet: which option worked better, and which one people would choose.4.Call the winner by the rule below.5.Say whether Prefer is stated or shown.Stated preference is a leading read; the predicted outcome names the behavior that would confirm it.6.No Prefer signal, no bet.Run Test Iteration with the Prefer question added.PreferUseCallWinsWins or tiesImplement is openTiesWinsMay go ahead, labeled "won on Use, preference tied"; the bet carries that riskWinsLosesA split: Refine Design or Test Iteration, never ImplementLosesAnyNot this optionDoesn't separate on eitherDoesn't separateNo winner: Test Iteration, or the business goal decides and the bet says soSize the bet to the evidenceBets happen at two levels. Strategic bets choose a direction the business should invest in. Tactical bets choose a version, flow, or message. Both use the same rule; what changes is how much evidence the bet needs.At the top of the triangle, bets shape interfaces and are cheap to undo. At the bottom, bets set strategic direction and are expensive to reverse. The deeper the bet, the stronger the evidence it needs, and the further up Lead's KPI chain its prediction has to reach.Tactical betsare judged on design KPIs, with a product KPI to watch.Strategic betsneed a credible line to business KPIs and the business goal before the team commits.When the cost of being wrong is high, reduce the risk before committing:Four Big Risks:check value, usability, feasibility, and business-model risk.Smoke test:measure real interest before building.Pilot review:run the bet with a small group before a full rollout.Proof of concept:confirm the hard part works before investing more.Ranking several betsSometimes more than one option clears the rule, and capacity allows only some of them. Prioritizing techniques rank bets that already have evidence. They come after the comparison, never in place of it.RICE:scores reach, impact, confidence, and effort.ICE:a faster version that scores impact, confidence, and ease.Weighted scoring:ranks options against criteria the team agrees on first.Value vs. Complexity:plots options by impact and effort.MoSCoW:sorts work into must, should, could, and won't.Eisenhower Matrix:separates what is urgent from what is important.Now / Next / Later:sequences bets over time without fixed dates.LNO:sorts tasks into leverage, neutral, and overhead.How to make a bet1. Name what is being chosenName the options and the variable that separates them. Say whether the bet is tactical or strategic, and which initiative it serves.2. Read the Use line and the Prefer lineBring the comparison readout. Which option worked better? Which would people choose? Were both measured the same way for every option?3. Apply the winner ruleCall it with the table above. If Prefer was never measured, stop here: the next move is Test Iteration with the Prefer question added.4. Name the costEvery bet gives something up: build effort, speed, an audience it serves less well, or a preference that tied. Say what it is.5. Write the predicted outcomeName the behavior that should follow, the metric that will show it, and when Lead: Results will check. If the Prefer signal was stated, the predicted behavior is what confirms it.6. Lock the owner and the next stepA bet is not finished until someone owns it and the outcome check is on the calendar.What goes into a betthe initiative it servesthe options that were compared and the variable between themthe Use line and the Prefer line, each measured the same way per optionthe winner-rule callthe cost the team acceptsthe business goal and the workflow that owns the resultthe predicted outcome: the behavior, the metric, and the windowthe owner and the next stepFrom bets to resultsA bet is a prediction written in Lead's language. It says that if the team backs this option, people will keep choosing it, and specific numbers will move.Every bet names four things that come from Lead:Business goal:the pressure it serves (Growth, Retention, User Experience, Efficiency, or Operations) and the specific goal.Workflow:the function that owns the move or feels the result.Expected outcome chain:the design KPI it should move, the product KPI that should follow, and the business KPI behind it.Outcome check:the baseline, the window, and when Results reads it.In Lead Results, work runs through four layers: Initiatives and Bets come from Focus, Findings from Measure, and Outcomes are what Results checks. When an outcome check shows the chain broke, that gap goes back to Measure as a new hunch, or to Focus to re-rank the bets.Where Bets fit in GlareInitiatives define what needs attention.Variables name what changes.Comparing repeats the measurement across the options.Bets commit to the option people prefer and the outcome it should bring.A single test that ends in Implement, Refine Design, or another decision type is a Findings next move, not a bet. Bets start when there are options to choose between.AI PromptThis prompt helps you turn a comparison finding into a named decision your team can act on.Start with a finding from a completed comparison, review, or research readout. It guides you to:Ground the decision in the strongest signal from the comparisonName the tradeoff the team is acceptingChoose exactly one of the five decision types: Implement, Refine Design, Test Iteration, Revisit Later, or Do Not PursueLock the next move so someone owns what happens after the meetingYou'll end with a clear decision backed by signal, with a tradeoff named and a next step recorded.Use this in any design review, sprint, or stakeholder readout where the team has evidence but keeps circling the same questions without committing.AI SkillThe Decisions skill file teaches your AI the full five-move taxonomy and decision-making process so it can help you convert any comparison finding into a named, recorded decision that downstream teams can act on.Load it when you need to go deeper on the criteria for each decision type, writing tradeoffs explicitly, or distinguishing strategic decisions about direction from tactical decisions about the next move on a specific concept. It gives your AI:The five tactical decision types with the criteria for choosing between themThe five-step process from naming what is being decided through locking the next moveThe guidance for grounding every decision in a signal rather than a preferenceThe decision record structure including the initiative, signal, tradeoff, type, and next stepDownload the skill file below to use the full Decisions framework with your AI assistant.

Related links

Tom Greever

Tom Greever's IDEAL framework (Identify, Describe, Empathize, Lock) for responding to design feedback, paired with four categories: business, design, research, limitations. Useful when a designer wants a script for hard stakeholder questions.

Bryan Zmijewski

Bryan Zmijewski breaks design decisions into six layers (strategic, conceptual, structural, aesthetic, functional, technical) so teams can move faster on each. Useful when teams confuse strategic and tactical design choices and slow down both.

Lia Garvin

Practical guide from Google Design on running design reviews that end in clear decisions. Useful when reviews drift and the team leaves without a call.

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