Back
Mobile Surveys: How to Design Them for Small Screens

Matthieu Saussaye

Most of your respondents will answer on a phone, held in one hand, between two other things. Design for that person first.
Mobile surveys are no longer a special case. StatCounter puts mobile at roughly 60% of worldwide web traffic, and on survey links shared by SMS, WhatsApp or social media the share is usually higher. A questionnaire built on a laptop and "made responsive" later loses people at every grid and every long text box. SmartInterview is the fastest way to run a study that works on a phone from the first question: the AI drafts a short, mobile-first questionnaire from your objective, and respondents can answer open questions by voice instead of typing.
The Short Answer
Assume mobile first. Most respondents will open your link on a phone, so test every screen on a small device before launch.
Keep it short. Aim for 5 to 8 minutes on mobile. The same questionnaire takes longer on a phone than on a desktop, and breakoff is higher.
One question per screen, big tap targets, no horizontal scrolling. Replace grids with item-by-item questions and long dropdowns with lists or search.
Favour closed questions that fit a thumb: single-select, short multi-select, 5-point scales, yes/no, NPS. Keep open questions few and focused.
Let people speak. Voice answers remove the main reason open answers are short on mobile: typing on a small keyboard.
Analyse by device. Add device type to your cross-tabs and check that phone and desktop respondents did not answer differently because of layout.
Mobile vs Desktop Surveys at a Glance
Desktop | Mobile | What it means for design | |
|---|---|---|---|
Screen | Wide, mouse pointer | Narrow (about 360 to 430 px wide), thumb | One column, one question per screen |
Completion time | Baseline | Longer for the same questionnaire | Cut questions, not just shrink fonts |
Breakoff | About 5% on average | About 13% on average (Mavletova and Couper meta-analysis) | Put key questions early, show progress |
Grids / matrices | Workable | Squeezed, more straight-lining | Split into separate questions or swipe cards |
Open text answers | Longer, typed on a keyboard | Shorter, typed with thumbs | Offer voice, add a follow-up when vague |
Context | Desk, focused | Commute, sofa, queue, interruptions | Save progress, allow resume, avoid timers |
Best question types | Almost all | Single-select, short multi-select, scales, yes/no, NPS, voice | Choose the format the thumb can do in one tap |
Breakoff figures come from the meta-analysis of mobile web surveys by Aigul Mavletova and Mick Couper, which found breakoff on smartphones several times higher than on PCs across published studies. The exact gap depends on how well the survey is optimised: a phone-friendly design closes much of it. For the basics of each format, see our survey questions guide.
What Changes When the Respondent Is on a Phone
What a mobile survey actually is
A mobile survey is any online survey completed on a smartphone or tablet. In practice there are three kinds: a web survey opened from a link (the most common, no app needed), an in-app survey shown inside a product, and an SMS survey where questions arrive as text messages. This article focuses on the first, because it is the one you control fully and the one most panels and customer lists use.
Why the same questions get different answers
The device changes behaviour in three measurable ways. First, time: reading, scrolling and tapping are slower on a small screen, so the same questionnaire runs longer. Second, effort: when a screen is crowded, people take shortcuts, such as picking the first visible option or giving the same rating down a whole grid (straight-lining). Third, open answers shrink, because thumb-typing is tiring. These are forms of survey bias caused by layout, not by opinions, and they are why response rates and data quality drop when surveys ignore phones.
Tap targets and readable text
Platform guidelines give clear minimums. Apple's Human Interface Guidelines recommend touch targets of at least 44 x 44 points; Google's Material Design recommends 48 x 48 dp; WCAG 2.2 sets 24 x 24 CSS pixels as the accessibility minimum (success criterion 2.5.8). For surveys, make the whole answer row tappable, not just the small radio button, and keep body text at 16 px or more so the phone does not zoom.
Worked example: how much sample to field
Say you need 400 completed mobile responses and expect a 13% breakoff among people who start. You need 400 / (1 - 0.13) = 460 starts. If your design is poor and breakoff climbs to 25%, you need 400 / 0.75 = 534 starts, 74 more people paid for or contacted to get the same result. Shortening the survey is often cheaper than buying the extra sample.
Worked example: a 7-minute time budget
A planning rule of thumb I use on mobile, not a published norm: about 10 seconds for a single-select or yes/no question, 15 to 20 seconds for a short multi-select or a scale with a sentence to read, and 45 to 60 seconds for an open question. A 7-minute budget (420 seconds) then fits, for example:
Block | Questions | Seconds each | Total |
|---|---|---|---|
Screeners (single-select) | 3 | 10 | 30 s |
Core closed questions | 10 | 15 | 150 s |
NPS + reason | 1 + 1 | 10 + 50 | 60 s |
Open questions (text or voice) | 2 | 60 | 120 s |
Demographics | 4 | 10 | 40 s |
Total | 21 | 400 s, about 6.7 min |
If the draft goes over budget, remove whole questions. Making each question smaller rarely saves as much time as cutting the ones you will never use in the analysis.
How to Build a Mobile Survey, Step by Step
1. Write the decision, then cut to it
Start from the one decision the results must support, and list only the questions that feed it. On mobile every question costs more, so "nice to know" items go first.
2. Use one question per screen
A single question with its answers should fit on one screen without scrolling. Mavletova and Couper found that scrolling within a short page can work, so the rule is not dogma, but one question per screen is the safest default and makes logic and progress easy to follow.
3. Pick thumb-friendly question types
Use these formats, written with full answer options:
Single-select (single-select questions): "How often do you buy plant-based yoghurt?" Weekly or more / 2 to 3 times a month / Once a month / Less often / Never.
Short multi-select (multiple-choice questions), 7 options at most: "Where did you buy it in the last month? Select all that apply." Supermarket / Discounter / Organic shop / Online / Other.
5-point scale (Likert scale), shown vertically: "The app is easy to use." Strongly disagree / Disagree / Neither / Agree / Strongly agree.
Yes/no (dichotomous questions): "Have you used a meal-delivery service in the last 3 months?" Yes / No.
NPS (Net Promoter Score): "How likely are you to recommend us to a friend or colleague?" 0 to 10, laid out in two rows (0 to 5, 6 to 10) if it does not fit one row.
Open question (open-ended questions): "What is the main reason for your score?" Text or voice.
4. Replace grids and long lists
A 6-statement agreement grid becomes six single questions, or swipeable cards with the same scale. Keep the scale labels identical across items. A dropdown of 200 countries becomes a type-to-search field. Ranking 8 items by drag and drop is hard with a thumb; ask for the top 3 instead (see rank order questions and matrix questions).
5. Plan open answers for thumbs or voice
Ask at most two or three open questions, each about one thing. Offer a voice answer: speaking is faster than thumb-typing and people say more. Then add a follow-up when the answer is vague, for example "too expensive" becomes "Too expensive compared with what you buy now, or for what it does?" More in voice surveys vs traditional surveys.
6. Test on real phones, then soft-launch
Open the survey on a small Android phone and an iPhone, on mobile data, in portrait. Time yourself. Then send it to 20 to 30 respondents, check the breakoff point and the median time, and fix the question where people drop.
7. How to do it in SmartInterview
In SmartInterview, you type your objective, for example "find out why trial users of our app do not subscribe", and the AI drafts the full questionnaire with a quality score, so you can see where it is too long or a question is leading before launch. Respondents come from your own customer list through a reusable link tagged by source, or from an audience recruited in Switzerland and Europe, with screen-outs and quotas enforced live. On the phone, respondents can answer open questions by text or voice, and when an answer is vague the AI asks one follow-up question. In our studies, AI follow-ups deliver up to 4x longer answers and 2.2x more topics. Open answers are then coded automatically into themes linked to the verbatims, with sentiment and cross-tabs, and you can export to Excel or SPSS. Results typically arrive in about 48 hours.
Where I draw the line
I do not run a 20-minute questionnaire on mobile and hope a responsive template will save it. If a study genuinely needs a long battery of grids, such as a detailed attribute rating of ten brands, I either split it into two short surveys or accept that it is a desktop study and recruit accordingly. Pretending otherwise produces straight-lined data that looks complete and means little.
The other line: I no longer accept three-word open answers as "what mobile respondents give you". That was a typing problem, not a respondent problem. When people can speak, and get one well-placed follow-up, the phone becomes the best device for open answers, not the worst.
Matthieu Saussaye, Founder of SmartInterview.
Mobile Survey Checklist
Objective written in one sentence, and every question mapped to it.
Estimated time under 8 minutes on a phone, measured, not guessed.
One question per screen, no horizontal scrolling, text at 16 px or more.
Tap targets of at least 44 x 44 points, with the whole answer row clickable.
No grids: each item is its own question or a swipe card.
Multi-select lists of 7 options or fewer, plus "Other" and "None of these".
Two or three open questions at most, with voice answers allowed.
Progress shown and the ability to resume after an interruption.
Tested on Android and iPhone, on mobile data, before a 20 to 30 person soft launch.
Device type recorded so you can cross-tab results by phone vs desktop in your analysis tool.
Common mistakes
Designing on a laptop and checking mobile last. The layout problems are found after fielding.
Keeping matrix questions because "the tool makes them responsive". Responsive grids still produce more straight-lining.
Long dropdowns for age, country or brand lists. Use ranges, search or short lists.
Horizontal 0 to 10 scales that run off the screen, so respondents cannot see the high end.
Too many required open questions, which drive breakoff and produce one-word answers.
Long intro pages and consent walls before the first question. Keep the intro to two lines and link to details.
Ignoring device in the analysis. A difference between segments can be a difference between devices. Check it before you report.
Frequently Asked Questions
What is a mobile survey?
A mobile survey is a questionnaire completed on a smartphone or tablet, usually a web survey opened from a link in an email, SMS, QR code or app. It uses the same questions as any online survey, but its layout, length and question types are designed for a small touch screen.
How long should a mobile survey be?
Aim for 5 to 8 minutes, which is usually 15 to 25 short questions. The same questionnaire takes longer on a phone than on a desktop, and breakoff rises with length, so cut questions rather than shrinking them.
Are mobile survey results as reliable as desktop results?
They can be, if the survey is designed for phones. Studies by Mavletova, Couper and others find most differences come from layout, such as grids and long text boxes, rather than from the people. Record device type and compare results by device to check.
Which question types work best on mobile?
Single-select, short multi-select lists, 5-point scales shown vertically, yes/no questions and NPS. Avoid large grids, long dropdowns and drag-and-drop ranking of many items.
Should I use matrix questions on mobile?
Generally no. Split a matrix into separate questions with the same scale, or use swipeable cards. If you must keep one, limit it to 3 or 4 items and a 5-point scale.
Can respondents answer a survey by voice on their phone?
Yes, if the survey tool supports it. In SmartInterview respondents can answer open questions by voice or text, and the AI asks a follow-up when an answer is vague, which gives longer and more specific answers than thumb-typing.
How do I get people to answer a mobile survey?
Send the link where people already are (SMS, WhatsApp, email opened on the phone, a QR code at the point of sale), say how long it takes, keep the first question easy, and offer an incentive where appropriate. Panels handle invitations and incentives for you.
Want a survey that works on a phone from the first question, with voice answers and coded themes at the end? SmartInterview is free for your first 50 respondents. See how Pharmalp used it to choose between four product concepts in the Pharmalp case study, or check the plans.



