User Experience Survey Template
A 15-question user experience survey you can send to real users this week: the ten System Usability Scale items exactly as they were published, then five questions that turn the score into a list of things to fix. Answer it below, lift the wording, or open the whole thing in the editor.
- 15questions
- About 4 minto complete
- 1 to 10are the System Usability Scale
- 0 to 100and not a percentage
Which usability score are you allowed to publish?
Guides on this topic name a standardised score and stop there. What decides whether you can use one is what it costs and what you may print. Here is the whole set, with each owner's own position.
| Instrument | What it costs | What you get | What you may do with it |
|---|---|---|---|
| System Usability Scale, 10 items | Nothing | One number from 0 to 100 for one interface | Free to use, industrial evaluation included, on one condition: a published report acknowledges the source. It carries "(c) Digital Equipment Corporation, 1986", so it is not free of copyright.[1] |
| SUPR-Q, 8 items | Questions free; the percentile database is a licence at 3,000 to 5,000 US dollars a year | A website quality score and a percentile rank against other sites | The owner permits the questions with a citation and sells the benchmark that gives the rank meaning.[6] |
| UEQ, and the 8-item UEQ-S | Nothing | Scores on several experience scales rather than one number | Free of charge; the site asks you to cite Laugwitz, Schrepp and Held (2008).[7] |
| QUIS | A licence fee | Interface satisfaction ratings across several areas | Licensed through the University of Maryland office that handles technology commercialisation.[8] |
| SUMI | A commercial licence | A standardised usability measure used in research | Commercial licensing requirements apply, and that is the usual reason teams pass on it.[9] |
| SEQ(R), 1 item | Nothing to ask it | One rating for one task, asked immediately after it | It became a registered trademark in December 2021. Name it; do not brand an item of your own with it.[5] |
| UMUX-Lite, 2 items | Nothing to ask it | A fast stand-in for a full scale score | Published in an ACM conference paper, and the permission printed on that series stops short of copying for commercial advantage. Described here, not printed.[10][11] |
Why the ten items are printed here in full. The scale was made freely available by its author, who asked one thing in return: that any published report acknowledges the source.[1] He named industrial evaluations in the same passage, so a template like this one sits inside the grant, and the copyright line travels with the items.
What the score is for. One number for one interface, from people who have just used it. It is deliberately coarse: it names no screen and was never meant to, which is why five more questions follow it here.
Inside an IT team the same thing is usually called a user satisfaction survey. It is the same instrument, pointed at a tool people were given rather than one they chose, and it is scored in exactly the same way.
All fifteen questions are already built in the survey creator. Change the five diagnostics to suit your product and leave the ten scale items alone.
If the thing being judged is not one interface, something else is closer. When the subject is the app itself, its store listing and what it does on the phone, the app feedback survey asks about that. When named testers are checking a release against scripted steps before it ships, that is the user acceptance testing survey. When the question is whether to buy or renew a tool rather than whether it works, the software evaluation survey scores it against criteria you set.
The 15 questions, and which ten are the score
Ten items you are allowed to reproduce, five that are ours, and a line on each one saying what a bad answer looks like.
The System Usability Scale, questions 1 to 10
The published scale, word for word and in its published order, on a 1 to 5 agree scale. It carries the line "(c) Digital Equipment Corporation, 1986" and the source to cite is Brooke (1996).[1] Odd and even items point in opposite directions on purpose, which is why they are scored differently.
- I think that I would like to use this system frequentlyPull rather than satisfaction. A low answer from someone who uses it daily is the one worth chasing.
- I found the system unnecessarily complexAgreement is the bad answer here. It is scored backwards, so a 5 contributes nothing to the total.
- I thought the system was easy to useThe plainest item on the scale, and usually the first one to move after a redesign lands.
- I think that I would need the support of a technical person to be able to use this systemAgreement means the interface needs a person attached to it. Expensive, and invisible in usage data.
- I found the various functions in this system were well integratedA low answer usually means two teams shipped two halves of one flow and neither owns the join.
- I thought there was too much inconsistency in this systemAgreement points at patterns rather than screens: two controls that do the same job differently.
- I would imagine that most people would learn to use this system very quicklyLearnability. A low answer beside a high ease answer means it is fine once somebody shows you.
- I found the system very cumbersome to useAgreement is the bad answer, and it separates slow from confusing better than item 3 does.
- I felt very confident using the systemLow confidence is the answer to read next to question 12 rather than on its own.
- I needed to learn a lot of things before I could get going with this systemAgreement is the bad answer. This is your onboarding score, from people who are already past onboarding.
Device and task, questions 11 to 13
Three questions the scale does not ask. Device is the split that explains most of the swing between two runs, and confidence separates a task that went well from one that only felt like it did.
- Which device were you using just now?Split every score by this before comparing anything. A phone result and a desktop result are two results.
- Were you able to finish what you came here to do?What they say they did, not what they did. A "no" beside high scale answers means they blamed themselves.
- How confident were you that you were on the right track?Answerable on either branch of question 12. Low confidence with a "yes" above it is a silent failure: they finished, and they got it wrong.
The fix, questions 14 and 15
One open box before the end, then a rating, because a survey that finishes on a text box loses its last answer. Anyone who skips the box can still answer the rating.
- What slowed you down or confused you most?The only item that names a screen. Tag each answer to one screen before you read any number.
- How much would fixing the biggest problem here improve it for you?Priority, and it does not lean on the optional box above. A low answer stops you building a fix nobody wanted.
About the words "this system". Brooke printed the items that way and this template keeps them. Swapping in your product name is an adaptation you are making, not part of the published scale, and no approval for it exists. Name the product once in the invitation and the items can stay as they are.
How ten answers become one number from 0 to 100
Two different rules and one multiplication. Get any of the three wrong and the result is not a scale score, however sensible the number looks.
| Step | The rule | Worked example |
|---|---|---|
| 1. Score the odd items | Questions 1, 3, 5, 7 and 9: the scale position minus 1. | Answers of 4, 4, 5, 4, 4 give 3 + 3 + 4 + 3 + 3 = 16 |
| 2. Score the even items | Questions 2, 4, 6, 8 and 10: 5 minus the scale position. | Answers of 2, 2, 1, 2, 2 give 3 + 3 + 4 + 3 + 3 = 16 |
| 3. Add the ten contributions | Each item contributes 0 to 4, so the sum runs from 0 to 40. | 16 + 16 = 32 |
| 4. Multiply by 2.5 | That is the whole scale: one number from 0 to 100. | 32 x 2.5 = 80 |
Why the odd and even items are scored differently. Half the items were written so the natural answer is agreement and half so it is disagreement, which makes people read each one instead of running down the page.[1] Scoring them alike would cancel that design out.
Score each person, then average the people. Work out one number per respondent and take the mean across respondents. Scores for individual items are not meaningful on their own, so never report question 4 as a finding.[1]
It looks like a percentage and it is not one. A 70 is not seventy per cent of anything.[2][3] The average across the 500 studies one consultancy has benchmarked is about 68, which puts a 70 a shade above ordinary.[3]
Rough reference points. Around 80 and above sits in the top tenth of studied systems; a 73 is only about the top third.[2] The adjective labels people quote come from Bangor, Kortum and Miller, and the bands attached to them circulate with no readable source, so none is printed here.[4]
The five diagnostics are never scored. Folding them in turns a comparable score into one that compares with nothing.
When to send it, and what to say before people answer
Three decisions this scale needs and no other survey does. The parts every send shares, the link, the reminder and the closing date, are set out in the survey research guide.
Wait until they have used it, then ask before anyone talks to them
The scale is administered after real use and ahead of any debriefing conversation. A discussion first changes the answers you get.[1]
Ask for a first reaction, and a middle mark when an item does not fit
That instruction ships with the scale: record the immediate response, tick every item, and use the centre point rather than leaving a blank.[1]
Give the same people a cool-off before you ask again
Asking the same users repeatedly is how a willing group stops answering, and a periodic read is what this scale is for.[12]
Reading the result, and what it cannot tell you
A number and a pile of comments, only useful read together, and the boundary of what a satisfaction measure can say at all.
Read the open box first, then the score. Tag every answer to one screen or one step before you look at a number. The score tells you how bad it is; the tags tell you where.
Rank the fixes by question 15, not by how often something is mentioned. Five people naming a small annoyance is not the same as two people saying that fixing one thing would change everything for them.
It measures satisfaction, and only satisfaction. Brooke sets out three classes of usability measure: whether people complete tasks, what that costs them in time and effort, and how they feel about it.[1] A survey is the third. It cannot tell you whether a task was finished or how long it took.
Question 12 is what people say, not what they did. People report finishing tasks they abandoned, and abandoning tasks they finished. Watching five people attempt the same task gives you the answer this survey structurally cannot.
A small sample moves on its own. Ten people give you a number that looks different next month with nothing changed. If the change you are testing is small, keep the wording and ask again next quarter rather than declare a win.
The number is not comparable across different things. A checkout flow and an admin console do not sit on one ladder, and neither do two firms measuring different tasks. Your own previous run, same items, same trigger and same device, is the only comparison you fully control.
On who sees the answers. Whoever owns the survey can open every response row, so promise that and nothing beyond it.
If the people using the interface are your own engineers, and the subject is the toolchain rather than one screen, the developer experience survey asks about builds and local setup instead.
User experience survey FAQ
How many questions should a user experience survey have?
The ten scale items are fixed: drop one and the score no longer compares with anyone else's. Around them, keep it small. Published guidance on this topic lands at five to ten questions, or at ten to fifteen. This template runs fifteen because ten of them are the scale.[13]
Should I use the System Usability Scale or write my own questions?
Both, and this template does both. The scale gives you one number to hold against your own last run and against published averages. Your own questions give you the list of things to change. Neither does the other job.
What exactly do I have to write down to use the scale?
There is no fee and no permission to request. The one condition its author set is that a published report acknowledges the source, so a line naming Brooke (1996) alongside the copyright notice is the whole obligation. Never describe the scale as public domain: it carries "(c) Digital Equipment Corporation, 1986".[1]
Can a survey replace usability testing?
No. A survey records what people say about using something. A test records what they did, where they stopped and how long it took. Run the survey to learn how bad it is and whether that is moving. Watch people to learn where.
Can I change "this system" to my product name?
You can, and plenty of teams do, but it becomes your adaptation rather than the published scale. No approval for a substituted wording exists, so a strict comparison with published averages assumes the original items. Naming the product once in the invitation keeps both.
Methods and sources
What this template is based on
Questions 1 to 10 are the System Usability Scale, from its author's published chapter; questions 11 to 15 were written here to give the fix list a score cannot.[1]
Question licensing
The scale is free to use, commercial work included, on one condition: a published report acknowledges the source. It is not public domain, and it carries the line "(c) Digital Equipment Corporation, 1986". SUPR-Q, UEQ, QUIS, SUMI, SEQ(R) and UMUX-Lite are named and described here, never reproduced; their rights are in the table above.[6][7][8][9][10]
Limits and disclosure
A survey measures satisfaction, one of three usability classes Brooke describes, and cannot measure task completion or time on task.[1] The published averages pool one consultancy's own studies, not a random sample of software.[3]
SuperSurvey makes the editor this template opens in. The scale and the questions are free to use with or without it.
References
- Brooke, J. "SUS: A quick and dirty usability scale." In Usability Evaluation in Industry. Taylor and Francis, 1996. ahrq.gov (PDF)
- Nielsen Norman Group. Measuring perceived usability. NN/g. nngroup.com
- Sauro, J. Measuring usability with the System Usability Scale. MeasuringU. measuringu.com
- Sauro, J. Interpreting a SUS score. MeasuringU. measuringu.com
- Sauro, J. The evolution of the Single Ease Question. MeasuringU. measuringu.com
- MeasuringU. SUPR-Q: Standardized User Experience Percentile Rank Questionnaire, product page. measuringu.com
- Laugwitz, B., Schrepp, M. and Held, T. User Experience Questionnaire, 2008. ueq-online.org
- Human-Computer Interaction Lab. Questionnaire for User Interaction Satisfaction. University of Maryland. cs.umd.edu
- Standardized usability questionnaires in healthcare: a 26-year scoping review. Frontiers in Digital Health, 2026. frontiersin.org
- Sauro, J. and Dumas, J. "Comparison of Three One-Question, Post-Task Usability Questionnaires." CHI 2009. ACM, 2009. measuringu.com (PDF)
- Sauro, J. The evolution of the UX-Lite. MeasuringU. measuringu.com
- Contentsquare. UX surveys, guide. contentsquare.com
- Maze. UX survey questions, guide. maze.co
Send the user experience survey this week
Fifteen questions, about four minutes, the ten scale items exactly as published. Open it in the editor, change the five that are yours to change, and share one link.
Use this template