Create Your First Signal

Product and design work often slows down when a decision has to be made across teams.

Say you have two ways a flow could work. Design likes one. Product likes the other. Engineering is worried about how hard it is to build. There is some feedback and a little data, but nothing points to a clear choice. The work keeps going. The decision keeps coming back. Confidence drops.

A design signal is how Glare breaks that loop. A signal is evidence that has been read and given a meaning. It says what users showed. It says if that supports or goes against what the team expected. A signal is not a feeling or a hunch. It comes from putting a hunch up against real evidence.

This guide walks one decision through the Decision Map, one block at a time. You will see where the signal comes from and what happens to it next.


Where to Start

First, ask one question: does the team already believe something about a certain concept?

  • No hunch yet: start in Define with Collecting. Learn first, so the hunch you write later starts from something real.

  • A hunch about a concept: start in Measure. If the test shows the team does not understand what is going on, go back to Define.

The example below starts at the very beginning, so you can see every block once.


The Example

A rewards team at a bank sees that new cardholders rarely use their points. Design wants to add a progress bar to the home screen. Product thinks people cannot find the rewards tab. Both could be right, and neither has evidence.


Define: Get the Situation Right

Define makes sure the team is solving the right problem for the right people. It happens before anyone tests an answer.

User Needs

Name what people are trying to do. New cardholders need to see what their points are worth. That is the Valuable need, in the Impact group.

Audience

Name whose evidence counts. Here it is customers in their first 90 days. It is not what the project team thinks they want.

Collecting

Gather what is already out there and decide how much to trust it. The team has rewards tab numbers (12% of weekly users visit the tab), notes from five interviews, support tickets, and evidence the Glare MCP brings back from earlier work. All of it goes into a Source Map. The map shows what the team knows, how sure it is and what is missing.

Situations

Read the Source Map and agree on what is happening: new cardholders rarely open the rewards tab, so earned points go unused. Situations is judged by credibility. Would new cardholders say this is what really happens to them?


Measure: Test One Hunch

Measure runs one test to get an answer. It always goes in the same order.

Concepts

Start with something people can react to. Points progress bar: shows how close each cardholder is to their next reward on the home screen, for new cardholders.

Hunches

Write what you believe about that concept, and why, in one sentence: We think people ignore rewards because points feel abstract. Then write the questions that would prove it right or wrong. Can participants say what 5,000 points buys? Do they notice the progress bar without help?

This is an external hunch, because it is about users. So it is checked with UX metrics. An internal hunch is held by a part of the business, like Marketing. It is checked with that team's own KPI. Product's hunch about finding the tab gets its own test. Two hunches, two tests. No one has to win the argument in the room.

Testing

Put the hunch up against real evidence. Write the test before it runs, and put the expected result in the sentence: Test the abstract-points hunch with 10 new cardholders using a concept test; expect 70% task success on the point value task. Setting the bar first keeps anyone from moving it later.

Findings

Make the call on the hunch first. Then read the evidence. With the current dashboard, 42% of people could say what their points buy. With the progress bar, 78% could. The hunch is supported.

Now write the signal: Low task success on the value question shows point value is unclear; supports the abstract-points test. Then decide how strong it is. It beat the expected result by a lot. But it came from one group of ten people. So it is moderate.

This is your first signal. It rests on evidence. It says what the evidence means. It says which way it pushes. Findings is judged by usefulness. Does the concept help people do what they came to do?


Focus: Compare and Place a Bet

One test tells you a concept helps. Focus tells you which option is strongest and if it is worth the commitment.

Initiatives

The signal goes to an initiative: Improve rewards visibility for new cardholders by end of Q1.

Variables

Name the one thing that changes. Here it is the version, which is the Iteration variable. One version is a progress bar toward the next reward. The other is the same bar with a dollar value. The metric, the test and the audience stay the same.

Comparing

Run the same test across the variable. An A/B test splits new cardholders in half for three weeks. The dollar-value version does better on task success.

Bets

Rank the options. Make the bet only as big as the evidence allows. Then write the bet: We bet the progress bar will move Drive adoption, affecting Product and Marketing; measured by rewards redemptions. Say what you gave up, too. The rewards tab redesign waits a quarter. Bets is judged by desirability. If users had the choice, would they pick this?


Lead: Show What It Drives

Lead picks up the bet and checks if it moved the business.

Business Goals

The bet helps Drive adoption. That is one of the 20 business goals, in the User Experience group.

Workflows

Name each part of the business that has to change. Product ships the dashboard change. Marketing adds the bar to the day 3 and day 10 welcome emails. Marketing also moves budget to pay for the design time.

Mapping

Link the design KPI to the product and business KPIs. Task success leads to more rewards tab visits. More tab visits lead to more redemptions.

Results

At the end of the set time, compare what happened with the target. Judge it met, partly met or missed. Results is judged by viability. Do people keep using it?


Share It with Glarify

Every block hands off one result. Glarify turns it into something one reader can use, in the place they already look. It never makes weak evidence look strong. The finding goes to the product lead as a short message. The first line says how strong it is. The bet goes to leaders as a one-page brief.


Make the Decision Hold

Once the work moves forward, the decision starts to drift. New people join. Details change. Questions come back. Keep the signal and the bet where everyone can see them. They are the shared reason for the choice. When questions come up, go back to three things:

  • What did we expect?

  • What did users show?

  • What did we decide because of it?

Set a watch item for anything that should reopen the decision, like redemptions dropping below the starting number. If that happens, the decision goes back for review. No one has to argue it from scratch.


Start Small

You do not need all 16 blocks on day one. Start with one decision the team is stuck on:

  • If there is no hunch yet, run Collecting and name the situation.

  • If there is a hunch, write the concept, then the hunch and its questions.

  • Write the test with its expected result, then run it.

  • Make the call on the hunch and write the signal.

  • Decide what moves forward.

Then do it again where it matters. Each signal makes the next decision easier. For any word in this guide, see the Glossary of Terms.

Related links

Sachin Rekhi

Sachin Rekhi on building a continuous feedback loop into your product. Useful when teams want fresh signal between releases.

Dr. Myriam Munezero

The article argues that outcomes only create value when they are explicitly tied to decisions, warning that teams often track outcomes without clarity on how those signals should change behavior or direction. Use this when diagnosing why outcome metrics aren’t influencing priorities and when deciding how to design work so signals clearly enable decisions.

Bryan Zmijewski

Bryan Zmijewski post on how consistent, measured design loops help teams build confidence and improve faster through customer feedback. Useful when a design team feels stuck and you want a frame to argue for tighter learning cycles.

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