UX KPIs · November 2026
What makes good UX measurable.
Ten metrics that prove the quality of a digital product: what they measure, what the benchmarks are – and what an ambitious UX improvement can realistically achieve.
| Metric | Measures | Benchmark | Ambitious target |
|---|---|---|---|
| Task success rate | Can users complete the task? | avg. 78% | ≥ 90% for core tasks |
| Time on task | How long does it take? | comparative only | core tasks much faster than before |
| Error rate | How often do things go wrong? | avg. 0.7 errors per task | core tasks almost error-free |
| SEQ | How easy does the task feel? | avg. 5.5 out of 7 | ≥ 6 |
| SUS | How good is overall usability? | avg. 68 out of 100 | ≥ 80 (top tier) |
| CSAT | How satisfied are users? | depends on scale | clear rise over your own baseline |
| Conversion rate | How many complete the goal? | industry-dependent | double-digit uplift |
| Drop-off rate | Where do users leave? | checkout avg. 70% | biggest drop-off step halved |
| Activation & retention | Do users arrive and stay? | activation avg. 37.5% | towards top-performer levels |
| Support contacts | What does confusion cost? | comparative only | far fewer tickets per task |
What is realistic?
75% on average – with a wide range.
- Across 99 case studies, UX redesigns improved their target metric by 75% on average in 2020. Half of the projects landed between 13% and 157%. NN/g
- In 2006–2008 the average was still 247%. The easy wins are gone – a double-digit gain today means good work.
- How much is possible depends mostly on the starting point: the weaker the product today, the bigger the jump.
Behaviour: what users do
-
01
Task success rate
The share of users who successfully complete a defined task – the most fundamental metric of all.
How it’s measuredsuccessful attempts ÷ all attempts × 100 – in usability tests, or in analytics where the goal is clearly measurable (e.g. a confirmation page).
BenchmarkAcross almost 1,200 tasks, the average is 78%. MeasuringU Where failure costs money or data, the target is 100%.
Ambitious improvementLift core tasks from below 78% to 90% or more. Instead of more than one in five users failing, only one in ten does – with a direct effect on completions and support costs. After an insurance app redesign in one of my projects, the app reached a task success rate of 95%.
-
02
Time on task
How long users need for a task – the measure of efficiency.
How it’s measuredTime from start to completion, usually for successful attempts only. Because a few very slow users skew the average, use the median or the geometric mean.
BenchmarkThere is no universal benchmark – time only means something in comparison: old vs. new version, your product vs. a competitor.
Ambitious improvementIn an older, well-documented redesign, one task dropped from 12 to 6 seconds and another from 75 to 15 seconds. NN/g For internal tools this is the most direct lever: every minute saved per task multiplies across all employees and all tasks.
-
03
Error rate
How often users slip up during a task: wrong field, wrong click, wrong input.
How it’s measuredtotal errors ÷ attempts, or as the share of attempts with at least one error.
BenchmarkAcross around 700 tasks, the average is 0.7 errors per task; about two in three users make at least one error. after MeasuringU
Ambitious improvementErrors are usually the reason behind long task times, failed tasks and poor ratings. MeasuringU Remove the most common sources of error and every other metric improves with it.
-
04
Drop-off rate
The share of users who start a process but don’t finish it – measured step by step.
How it’s measured1 − (completions ÷ starts), ideally for each step of a funnel.
BenchmarkIn e-commerce, 70% of carts are abandoned on average. A typical checkout shows 23.5 form elements; 12 to 14 would do. Baymard
Ambitious improvementFind the step with the biggest drop-off and simplify it, instead of rebuilding the whole process. Fewer fields, no forced account, clear error messages are the proven levers.
Perception: what users feel
-
05
Single Ease Question (SEQ)
One question right after a task: “How easy or difficult was that?” – on a scale of 1 to 7.
BenchmarkAcross 400 tasks and 10,000 users, the average is around 5.5. MeasuringU
Ambitious improvementScores above 6 show a genuinely easy task. The advantage: the SEQ shows exactly which task is the problem – faster and cheaper than any large survey.
-
06
System Usability Scale (SUS)
A standardised ten-item questionnaire that measures the perceived usability of a whole product on a scale of 0 to 100.
BenchmarkThe average across 500 studies is 68 – a “C”, not a percentage. MeasuringU At 80, a product beats almost 90% of all products measured.
Ambitious improvementFrom 68 to 80 points: from the middle of the pack to the top tier. Products scoring around 80 typically reach completion rates close to 100%. MeasuringU
-
07
User satisfaction (CSAT)
How satisfied users are with a product or a specific interaction – usually on a scale of 1 to 5.
How it’s measuredshare of 4 and 5 ratings ÷ all responses × 100 (“top-2-box”). Fix the scale and the calculation and never change them – otherwise the numbers can’t be compared. Sierra
BenchmarkThere is no universal benchmark. The comparison that matters most is with your own baseline.
Ambitious improvementAfter redesigning an insurance app and simplifying its sign-up, user satisfaction rose by 63% in one of my projects.
Business: what it delivers
-
08
Conversion rate
The share of visitors who reach the goal: purchase, enquiry, sign-up, application.
How it’s measuredcompletions ÷ visits × 100 – with a fixed definition of what counts as a completion.
BenchmarkHighly industry-dependent. Compare yourself over time, not with averages from other markets.
Ambitious improvementA study commissioned by Google found 8.4% more retail conversions with a 0.1-second faster mobile load – a correlation, not proof. web.dev After redesigning a shipping app, conversion rose by 17% in one of my projects.
-
09
Activation & retention
How many new users experience real value for the first time (activation) – and how many come back (retention).
BenchmarkOn average, 37.5% of new sign-ups reach the moment where a SaaS product delivers value. Userpilot 2025 After three months, a median of 2.5% of B2B users stay active – 15.6% for the best products. Amplitude 2025 (vendor data)
Ambitious improvementThe gap between median and top is more than sixfold. The lever is almost always onboarding: faster to first value, less at once.
-
10
Support contacts
Calls, tickets and chats that only exist because something in the product is unclear.
How it’s measuredsupport contacts ÷ transactions, broken down by topic – so you can see which part of the product creates cost.
Ambitious improvementAn information architecture redesign at Mozilla cut new help requests by 70%. NN/g For companies, this is often the UX metric that translates most easily into money.
What about the Net Promoter Score?
Popular, but overrated.
The NPS asks whether users would recommend a product. It’s simple – but weak as a UX metric.
NPS predicts growth.
A Stanford-led study found that NPS does not predict revenue growth. Lundmark et al. 2023
NPS beats satisfaction.
Where both were compared, satisfaction did as well or better. MeasuringU
How to use these metrics. Pick three to five that fit your goal – not all ten. A proven structure is Google’s HEART framework: Happiness, Engagement, Adoption, Retention and Task success. Rodden et al., Google Measure before the redesign, or you won’t be able to prove the impact afterwards.
About the numbers. Benchmarks come from large benchmark databases and are a guide, not a standard. The “ambitious targets” in the table are my assessment based on these sources. As of 19 November 2026.
Where do your numbers stand?
Let’s measure.
30 minutes. Your metrics. An honest view of where the biggest lever is.
Or write directly: me@marcelhoehne.com