A UX metric is a number that measures the experience people have with a product. Each one is tagged three ways: what kind of measure it is, whether it predicts or confirms change, and how far up the business it reports.
For example, when SecondBank tested its rewards dashboard, only 42% of new cardholders could say what their points were worth. With a points progress bar, that rose to 78%. Task success is the UX metric; 42% and 78% are its values.
Why it matters
Without agreed metrics, teams judge design by taste, and every review turns into a debate about opinions. Leaders hear that a change "tested well" and have no way to compare it with anything else they're funding.
Tagged metrics give everyone a shared scoreboard. They show whether people can actually use what was built, whether that early movement is likely to show up in product numbers, and how it connects to the business goals leaders already track.
Pick a small number of metrics before testing, record each value with its sample and date, and tag it so it can be traced up from design to product to business.
How we write one
Every UX metric is written as one sentence that names the metric, its value and its three tags:
A metric = a value (a type, a role, a level).
Task success = 42% (Behavioral, Leading indicator, Design KPI).
Writing it this way puts the number first and the tags beside it, so anyone can see what kind of evidence it is and where it reports without reading further.
How it differs from similar ideas
-
An observation is a specific moment someone saw, such as people scrolling past the balance. A UX metric is a measured value across a group, such as 42% task success.
-
A signal says what the evidence means for a test. A UX metric is part of the evidence; the signal interprets it.
-
A benchmark is the comparison point, such as last quarter's 12% tab visits. A UX metric is the current measurement being compared.
-
An outcome is a change in a metric over time, such as task success moving from 42% to 78%. A UX metric is a single measured value.
-
A business goal is what the business is trying to achieve. A UX metric at the Business KPI level reports toward it but is never the goal itself.
A good UX metric
-
Measures behavior, attitude or performance people can feel
-
Has a value, a sample size and a date
-
Names the instrument used, such as SUS, SEQ or task success
-
Carries all three tags: type, role and level
-
Uses a rate or ratio so it can be compared
-
Maps up the chain toward a business goal
What happens to it
-
Chosen. The team picks it when defining a test, before any data is collected.
-
Measured. A value is captured during testing, an experiment or from analytics, with a sample and a date.
-
Captured. It's added to Findings in Measure, or repeated across a variable in Comparing, in Focus.
-
Compared. It's set against a benchmark or an earlier value to show whether it moved.
-
Reported up. It informs bets in Focus, and in Lead it maps through user, product and business outcomes toward a business goal.
For Ray
This part tells Ray how to find UX metrics in an input, such as a test readout, an analytics export, a survey summary, an experiment report or a dashboard screenshot, and how to store them.
How to find one
-
Mark measured values. Read the input and mark every number that describes how people experienced the product: rates, scores, times, counts.
-
Name the metric and instrument. Record what was measured and how, such as task success, SUS, SEQ, NPS, time on task or tab visits.
-
Capture value, sample and date. Pull out the value, how many people it came from and when it was measured. If the sample is missing, leave it blank.
-
Tag the type. Attitudinal for what people say or feel, Behavioral for what they do, Performance for speed, errors or effort, Intelligence for model or system quality.
-
Tag the role. Leading indicator if it predicts later change, Lagging indicator if it confirms change after the fact, Proxy metric if it stands in for something hard to measure.
-
Tag the level. Design KPI for experience measures, Product KPI for product performance, Business KPI for business performance.
-
Write the sentence. Use the format: [Metric] = [value] ([type], [role], [level]).
-
Fill the record, score it, link it and store it using the rules below.
Cues to look for
-
Percentages, scores and times tied to a task or screen: "42% completed", "SUS of 68", "average 40 seconds"
-
Named instruments: SUS, SEQ, NPS, CES, NASA-TLX, task success, time on task, first click
-
Analytics columns such as visits, conversion, retention or redemption rate
-
Before and after pairs in a readout: "rose from 42% to 78%"
What doesn't count
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>If the statement...</p></th><th colspan="1" rowspan="1"><p>Store it as</p></th></tr><tr><td colspan="1" rowspan="1"><p>Describes a single moment or quote from a session</p></td><td colspan="1" rowspan="1"><p>Observation</p></td></tr><tr><td colspan="1" rowspan="1"><p>Explains what a number means for a test</p></td><td colspan="1" rowspan="1"><p>Signal, linked to the metric</p></td></tr><tr><td colspan="1" rowspan="1"><p>Sets a comparison point, such as last year's baseline or a competitor figure</p></td><td colspan="1" rowspan="1"><p>Benchmark</p></td></tr><tr><td colspan="1" rowspan="1"><p>Reports a change from one value to another after work shipped</p></td><td colspan="1" rowspan="1"><p>User, Product or Business Outcome</p></td></tr><tr><td colspan="1" rowspan="1"><p>States a target the business wants to hit</p></td><td colspan="1" rowspan="1"><p>Business Intent or Result target</p></td></tr></tbody></table>
The record
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Field</p></th><th colspan="1" rowspan="1"><p>What to fill in</p></th><th colspan="1" rowspan="1"><p>Values</p></th></tr><tr><td colspan="1" rowspan="1"><p>Statement</p></td><td colspan="1" rowspan="1"><p>The one-sentence metric</p></td><td colspan="1" rowspan="1"><p>[Metric] = [value] ([type], [role], [level]).</p></td></tr><tr><td colspan="1" rowspan="1"><p>Metric</p></td><td colspan="1" rowspan="1"><p>What was measured</p></td><td colspan="1" rowspan="1"><p>Short name, such as task success</p></td></tr><tr><td colspan="1" rowspan="1"><p>Value</p></td><td colspan="1" rowspan="1"><p>The measured value</p></td><td colspan="1" rowspan="1"><p>Number with unit</p></td></tr><tr><td colspan="1" rowspan="1"><p>Instrument</p></td><td colspan="1" rowspan="1"><p>How it was measured</p></td><td colspan="1" rowspan="1"><p>SUS, SEQ, NPS, CES, NASA-TLX, task success, time on task, analytics, other</p></td></tr><tr><td colspan="1" rowspan="1"><p>Sample size</p></td><td colspan="1" rowspan="1"><p>How many people or sessions it came from</p></td><td colspan="1" rowspan="1"><p>Number, or blank</p></td></tr><tr><td colspan="1" rowspan="1"><p>Date</p></td><td colspan="1" rowspan="1"><p>When it was measured</p></td><td colspan="1" rowspan="1"><p>Date or period</p></td></tr><tr><td colspan="1" rowspan="1"><p>Type</p></td><td colspan="1" rowspan="1"><p>What kind of measure it is</p></td><td colspan="1" rowspan="1"><p>Attitudinal, Behavioral, Performance, Intelligence</p></td></tr><tr><td colspan="1" rowspan="1"><p>Role</p></td><td colspan="1" rowspan="1"><p>How it relates to change</p></td><td colspan="1" rowspan="1"><p>Leading indicator, Lagging indicator, Proxy metric</p></td></tr><tr><td colspan="1" rowspan="1"><p>Level</p></td><td colspan="1" rowspan="1"><p>How far up it reports</p></td><td colspan="1" rowspan="1"><p>Design KPI, Product KPI, Business KPI</p></td></tr><tr><td colspan="1" rowspan="1"><p>Audience and context</p></td><td colspan="1" rowspan="1"><p>Whose experience and which test, concept or method</p></td><td colspan="1" rowspan="1"><p>Links to Audience, Test, Concept or Method</p></td></tr><tr><td colspan="1" rowspan="1"><p>Source</p></td><td colspan="1" rowspan="1"><p>The input and exact table, slide or passage</p></td><td colspan="1" rowspan="1"><p>Link to Input</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>How solid the measurement is</p></td><td colspan="1" rowspan="1"><p>High, medium, low</p></td></tr></tbody></table>
Store each value as its own metric record. A before and after pair is two records, and the change between them becomes an outcome.
Scoring confidence
Confidence measures how solid the measurement is, not whether the number looks good.
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Level</p></th><th colspan="1" rowspan="1"><p>When to use it</p></th></tr><tr><td colspan="1" rowspan="1"><p>High</p></td><td colspan="1" rowspan="1"><p>From a named instrument or analytics, with value, sample size, date and all three tags clear</p></td></tr><tr><td colspan="1" rowspan="1"><p>Medium</p></td><td colspan="1" rowspan="1"><p>Value and instrument are clear, but the sample, date or one tag has to be left blank</p></td></tr><tr><td colspan="1" rowspan="1"><p>Low</p></td><td colspan="1" rowspan="1"><p>Quoted secondhand, rounded loosely, or measured with an unnamed or unclear method</p></td></tr></tbody></table>
If any part of the record can't be filled from the source, leave it blank and lower confidence by one level. Never guess a tag; leave it blank and send it to review.
Linking
-
Check for duplicates first. If an existing metric has the same name, value, audience and date, add the new source to it instead of creating another.
-
Link to its Findings, test or method, and to its source input.
-
Link to a benchmark when one exists for the same metric, so the comparison is visible.
-
Link to the bets it informs and to the outcome it feeds, following the chain: Design KPI to User Outcome, Product KPI to Product Outcome, Business KPI to Business Outcome.
-
Note proxies. When a metric is tagged Proxy, link it to what it stands in for.
When to store and when to ask
-
Store automatically when confidence is high.
-
Add to the review digest when confidence is medium or low, when a tag can't be set from the source, or when the value contradicts an open test's expected result.
When to flag
-
A test or bet has no UX metric linked to it
-
A metric has no benchmark or earlier value to compare against
-
A Design KPI has no path up to a product or business KPI
-
A metric is reported without a sample size
-
Attitudinal and behavioral metrics for the same concept point in opposite directions
Worked example
Input: a test readout slide from the SecondBank rewards value test.
"Q: Can you say what 5,000 points buys? Current dashboard: 42% task success (n=10). Progress bar concept: 78% task success (n=10). Sessions run Oct 6 to 8."
What Ray finds:
-
"Task success" is the metric and the instrument, measured on the test question about 5,000 points.
-
"42%" and "78%" are two values, so they become two records, each with n=10 and the October dates.
-
Task success tracks what people do, so the type is Behavioral; it predicts later adoption, so the role is Leading indicator; it measures experience, so the level is Design KPI.
-
The move from 42% to 78% is noted for the User Outcome it may later support, and the two records are linked to the round 1 Findings.
Stored record:
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Field</p></th><th colspan="1" rowspan="1"><p>Value</p></th></tr><tr><td colspan="1" rowspan="1"><p>Statement</p></td><td colspan="1" rowspan="1"><p>Task success = 42% (Behavioral, Leading indicator, Design KPI).</p></td></tr><tr><td colspan="1" rowspan="1"><p>Metric</p></td><td colspan="1" rowspan="1"><p>Task success: can say what 5,000 points buys</p></td></tr><tr><td colspan="1" rowspan="1"><p>Value</p></td><td colspan="1" rowspan="1"><p>42%</p></td></tr><tr><td colspan="1" rowspan="1"><p>Instrument</p></td><td colspan="1" rowspan="1"><p>Task success</p></td></tr><tr><td colspan="1" rowspan="1"><p>Sample size</p></td><td colspan="1" rowspan="1"><p>10</p></td></tr><tr><td colspan="1" rowspan="1"><p>Date</p></td><td colspan="1" rowspan="1"><p>October 6 to 8, 2026</p></td></tr><tr><td colspan="1" rowspan="1"><p>Type</p></td><td colspan="1" rowspan="1"><p>Behavioral</p></td></tr><tr><td colspan="1" rowspan="1"><p>Role</p></td><td colspan="1" rowspan="1"><p>Leading indicator</p></td></tr><tr><td colspan="1" rowspan="1"><p>Level</p></td><td colspan="1" rowspan="1"><p>Design KPI</p></td></tr><tr><td colspan="1" rowspan="1"><p>Audience and context</p></td><td colspan="1" rowspan="1"><p>Participants: new cardholders; abstract-points test; current dashboard</p></td></tr><tr><td colspan="1" rowspan="1"><p>Source</p></td><td colspan="1" rowspan="1"><p>Test readout, slide 4, October 2026</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>High: named instrument with value, sample, date and clear tags</p></td></tr></tbody></table>
Next step: stored automatically with its 78% companion record, both linked to the round 1 Findings and to the 70% expected result on the test.
Edge cases
-
Tags not stated. Most inputs never say "Leading indicator". Set the tag when the metric's nature makes it clear; when it doesn't, leave it blank and send it to review.
-
Same metric, different levels. Rewards tab visits can be a Design KPI in a test and a Product KPI in analytics. Store them as separate records with their own level.
-
Tiny samples. A percentage from three people is still a metric. Store it with the sample size so signals can judge its strength.
-
Vanity numbers. Page views or downloads with no link to a need or goal can still be stored, but flag them when they're the only metric behind a decision.