Online surveys are a quick and incredibly useful tool for gathering all sorts of user feedback. In next to no time you can whip something up and start gathering valuable insights from real users. Often implementing the survey is the easy bit, it’s designing the damn thing that’s the tricky part. If I had a pound (or dollar) for every badly designed survey that I was asked to fill out, well I wouldn’t necessarily be a rich man, but I’d still be on to a nice little earner. By not asking the right questions, in the right order, and in the right way, you risk your survey generating a pile of data garbage, rather than the useful feedback and insights that you were hoping for.
Paul Simon sang about 50 ways to leave your lover. Well, I couldn’t quite stretch to 50, but here are over 40 ways to improve the UX of your online surveys.
1. Plan out what you need to capture
A good survey starts with good planning. List out all the insights and feedback that you want to capture, and think about how best to do this. Is an online survey even the best approach? Surveys are great for capturing quantitative data (i.e. ratings and numbers), but not so good for qualitative data. For example, rather than a survey, perhaps you’d get more insights from interviewing users instead.
2. Identify the survey audience
Part of your survey planning should involve identifying the audience for your survey, along with the device they’re likely to be using to fill it out. For example, if your audience is likely to be on a mobile, you’ll need to make sure the survey works well on a small screen. Also consider any qualifying questions you might need to ask to weed out any would be participants that you don’t want to include.
3. Consider how many responses you need
Having identified the survey audience, you’ll also need to consider how many responses you require and whether you’re likely to need to offer some sort of incentive. If you have a very large potential audience, then you might want to plan a sample size. Response rates will of course vary and a good tactic can be to send out staggered survey invites (either onsite invitations, or via email) which will allow you to monitor response rates and adjust accordingly.
4. Capture a mixture of quantitative and qualitative data
Surveys are ideal for capturing quantitative data, but often this will only tell you the ‘what’, not the ‘why’. It’s usually a good idea to capture a mixture of quantitative and qualitative data. For example, you might ask users to rate something and then ask them why they gave their rating. To reduce the overhead of analysing qualitative answers it can be useful to only ask for an explanation when an interesting answer has been entered. For example, a user has selected a high or low rating.
5. Don’t ask what your analytics will tell you
Using analytics tools such as Google Analytics you can capture a wealth of information about your users, without even having to ask them any questions. If your analytics can provide the insights then there’s no need to include these questions in your online survey.
6. Choose the right survey tool
There are lots of free survey tools available such as Google forms, SurveyMonkey, SurveyGizmo, SurveyTownand Typeform. However, if you want to be able to style your survey, include logic such as skipping questions or sections, or use fancy question types, then you’ll probably have to pay. Free survey tools also often limit the number of responses and number of questions per survey.
7. Focus on time to complete, rather than number of questions
Rather than aiming for no more than 5 or 10 questions, a better consideration is how long a survey is likely to take to complete. After all, 3 easy to answer questions might be quicker to answer than one complicated one. Whilst some survey tools will provide an approximation for time to complete, piloting a survey will give you a true indication.
8. Keep surveys as quick to complete as possible
Admit it. No one likes filling out surveys, so keep yours as short and sweet as possible. If you don’t truly need to ask a question, don’t. As a general rule the longer a survey takes to fill out, the higher the drop off rate. Think about what the minimum you can get away with asking is and how you can make your survey as painless to fill out as possible.
9. Have a logical flow
Like a good film or novel, your survey should have a logical flow. Avoid continually jumping from topic to topic and try to cluster related questions together, either in sections, or on pages. A good starting point for a survey is usually to capture information about the participant. This also allows qualifying criteria to be captured upfront so that would be participants that shouldn’t be filling out the survey can be excluded.
10. Keep short surveys to one page & break up longer ones
If a survey is quite short, for example 5 questions or less then try to keep it on the one page. Longer surveys should be broken up into multiple pages so that participants can tackle the survey a section at a time. Labelling pages is usually a good idea, for example ‘About you’.
11. Selectively ask for extra information
A good way to minimise the time it takes to complete a survey is to selectively ask for extra information. For example, if participants are asked to rate something, then you might only ask for why a rating has been given if a very high, or very low rating has been entered.
12. Front load with the most important questions
Along with the certainty of life, death and taxes, there is the certainty that participants will drop off your survey. As I’ve already mentioned, the longer the survey, the higher the chance of drop off. This is why it’s a good idea to front load a survey with the most important questions. This helps to ensure that you’ll at least get answers to these key questions.
13. Require only key questions
A survey is always about striking a balance between capturing the data you want, whilst minimising the work of participants. You should therefore require participants to answer only very key questions. If a question is not key, then either make it optional, or don’t include it in the first place.
14. Spend time on writing a good intro
A good introduction can make or break a survey. If you have a very poor response rate and therefore very few participants, it’s probably because the introduction is not doing its job very well. A good survey introduction should outline why the survey is being run, set expectations, such as time to complete, outline any incentives and of course sell the survey to would be participants. Keep introductions relatively short and make sure that you answer the all-important question, “Why should somone fill out this survey?”.
15. Say thank you and spell out any next steps
If someone has invested some of their own time and energy completing your survey, the least you can do is say thank you. Always thank participants at the end of a survey and if there are any next steps, such as results being made available, or a prize draw being made, outline when these will be and how participants can stay informed. If someone has been excluded from the survey because he or she does not meet the qualifying criteria, you should still thank them for participating. As I’m sure you’ve been reminded countless times, good manners cost nothing…
Choosing question types
16. Use open questions sparingly
Surveys are generally more suited to closed questions. For example, choosing an answer from a list, or selecting from a scale. Open-end questions can be useful, but can take a lot more time to analyse and for participants to answer. A good compromise is often to ask closed questions, with an optional open question to allow participants to provide more information. For example, why he or she selected that answer.
17. Choose appropriate question types
Survey tools increasingly support a bewildering number of different question types. From checkbox grids, to dropdowns, image selection, star ratings and likert scales; the choices are many. With so many different question types to choose from, which should you use? A good rule of thumb is to keep it as simple as possible. Basic radio button selections and likert scales are often all that’s required. For help choosing question types take a look at SurveyGizmo’s excellent advice for choosing survey question types along with their downloadable quantitative question guide.
18. Ask for recognition over recall
Along with keeping it simple, another good rule of thumb for choosing question types is to ask participants to select their answer from a list, rather than enter it using free text. Of course, it’s not always possible, or desirable to list all possible answers. If this is the case then don’t forget to provide a ‘Other’ option where participants can enter a different answer.
19. Use radio selections for small lists, drop downs for long lists
When presenting lists of possible answers use radio selections for small lists (e.g. 10 or less) and drop downs or auto-lookups (i.e. a participant starts typing and matches are shown) for longer lists.
20. Don’t present too many options
Try to avoid presenting participants with too options to choose from as this can make filling out a survey a long and painful process. Consider using auto-lookups to help standardise free text entry. For example, when entering a country of residence or profession.
21. Ensure that single selection choices are mutually exclusive
If you’re presenting a list of possible answers, consider whether they will be mutually exclusive or not. In other words, should participants be able to select more than one answer? For example, a participant will only ever have one title (‘Mr’, ‘Mrs’, ‘Miss’ etc…) but might have multiple mobile phones (e.g. work and personal phone). It’s generally best to use radio buttons or a dropdown for single selection, and checkboxes where multiple selections might be possible.
22. Provide a middle point for Likert scales
When asking participants to select from a Likert scale, provide a middle point where possible. This will generally mean using a 3, 5, 7, 9 or 11 point scale (i.e. 0–10). A middle point allows those participants that are neither positive, nor negative to select a valid answer, rather than having to gravitate towards one or the other.
23. Avoid radio button grids for mobile
Radio button grids, such as a Likert scale rating for multiple items can work very well on a big screen, but can be devilishly difficult to fill out on a small screen. If a lot of your participants are likely to be filling out the survey on their mobile try to avoid radio button grids. Instead, break the rows up into separate questions.
24. Use pictures where possible
Pictures can be a great way to aid clarity and comprehension within a survey. For example, using screenshots to help communicate what a question is asking about, or even to present the possible answers to a question.
Writing questions and options
25. Aim for neutral questions
There is a real art to writing good survey questions. They should be clear, easy to understand and as neutral as possible. Try to avoid introducing any unintentional bias due to the way that a question is framed. For example, rather than asking, “What do you really love about product X” you might ask, “How would you rate product X?”. SurveyTown have some examples of biased survey questions to avoid.
26. Don’t expect people to read instructions
People don’t read instructions. It’s true when it comes to everyday products, and it’s certainly true when it comes to surveys. This is why it’s important to make questions as self-contained as possible. Rather than providing instructions in the introduction, or at the top of a page, be sure to provide any additional instructions as part of a question, or directly below the question text.
27. Keep questions and instructions succinct
Rather than reading all of your finely crafted text word by word, participants are likely to scan the text of an online survey. It’s therefore important that questions and any instructional text are kept short and succinct. It can also be useful to highlight keywords and instructions, for example by making them bold.
28. Only ask one question at a time
It can be tempting to combine questions so that a survey appears shorter. For example, “Please rate the visuals and functionality”. The problem with doing this is that participants don’t know which part of a question to answer. It’s important to only ask one question at a time, so this example would be better phrased as 2 questions — “Please rate the visuals” and, “Please rate the functionality”.
29. Use plain, clear and easy to understand language
Always use plain, unambiguous language for questions. Generally, the simpler the language, the better. If a question is unclear then you’re likely to get participants answering different interpretations. A good way to ensure that questions are clear and easy to understand is to run through the questions with someone to find out how he or she interprets them.
30. Provide examples to help comprehension
It can often be useful to provide examples, so that participants have an indication of the type of answer you’re looking for. However, be careful about unintentionally introducing bias with the examples used. It’s often best to provide a number of examples so that participants don’t get fixated on just the one.
31. Randomize options where possible to avoid bias
Where you’re asking participants to select from a list, or from a grid, it’s often a good idea to randomise the order of that list in order to reduce bias. However, if the list has a usual order, such as numerical or alphabetical then you obviously shouldn’t randomise it.
32. Allow for ‘Other’ or ‘N/A’ options
Rather than forcing participants to select at least one item, always consider if you’ll need to provide a ‘Other’ or not applicable (i.e. ‘N/A’) option. When a ‘Other’ option is provided it’s also a good idea to capture details via a text box, such as the example below.
33. Size text boxes for expected input
Participants will get an initial idea of the amount of text to enter, based on the size of a text box. Therefore, it’s important to ensure that text boxes are sized appropriately for the required content. For example, an email text box should be large enough to easily accommodate a long email address, and if you expect a participant to enter lots of free text, don’t just provide a narrow one line text box.
34. Validate answers where possible
Where possible it’s a good idea to validate text based answers within a survey. For example, checking that an email address is in the appropriate format, or that a valid date has been entered for a date of birth.
35. Use branding to build trust
It’s important that participants trust the survey put in front of them, otherwise they’re unlikely to complete it. Aside from creating a professional and well-thought out survey, branding can really help to build trust. It’s a good idea to use your company or product branding for a survey and to generally follow good UI practice, such as adequate sized text, harmonise colours and good contrast.
36. Emphasise question text
The question text should be the focus for a survey, so it’s a good to make it stand out on the page. For example, by increasing the font size, or perhaps by making it bold.
37. Show progress for longer survey
For surveys involving multiple pages, it’s a good idea to include a progress indicator. For example, the percentage complete. This lets participants know how far into the survey they are and how much still remains.
38. Use question numbers for longer surveys
For longer surveys featuring multiple pages it can be useful to show question numbers. This gives participants an idea of how much of the survey they have answered. For short surveys of 1 or 2 pages, question numbers are best avoided.
39. Check styling and usability across devices
Gone are the days when you just had to check that an online survey works well on a desktop browser. With mobile and tablet traffic soaring, you should also check that a survey works well on small screens, and on touch screens. Large question grids are best avoided if you’re likely to get a lot of mobile traffic, and you might even consider creating a mobile specific version if the survey tool is not very responsive.
40. Pilot surveys & tweak accordingly
Like a once sick hedgehog that you’ve lovingly nursed back to full health, once a survey is out in the wild it’s very hard to bring it back into captivity. This is why it’s such a good idea to pilot a survey before publishing. This will provide an opportunity to find out if questions are being interpreted as you expect and to generally see whether the survey needs tweaking before being made available to a wider audience.
41. Thoroughly test any question logic
You usually only have one shot at getting a survey right. Make a mistake with the question logic, such as accidently excluding the wrong set of participants, or using the wrong question skip logic and your precious survey data could be in ruins. This is why it’s really important to thoroughly test any question logic before publishing a survey.
42. Spend time on the invitation
I shouldn’t think that when you wake up, your first thought of the day is, “Oh, I hope I get to fill out a survey today”. Surveys are not exactly top of most people’s list of things to do, so it’s important that a survey invitation stands out and includes a clear call to action. If you’re sending out an email invitation pay particular attention to the subject line and of course the text of the email. As with a good survey introduction keep the invitation relatively short and make sure that you answer the all-important question, “Why should somone fill out this survey?”. For help writing a good subject line check out SnapSurveys 7 Tips for Writing Great Email Subject Lines.
43. A/B test different invitations
A good way to determine the best invitation for a survey is to A/B test different variations. For example, you might use different subject lines, different emails, or even different incentives to find out which leads to the higher completion rate.
44. Send out reminders
With emailed survey invitations you tend to get an initial spike of activity, as people act on the email, and then completions gradually tail off, until you get very few after a 3–4 days. A good way to increase completion rates is to send out reminders to anyone who has yet to complete the survey. Allow about a week before sending out a reminder so that you’re not bombarding would be participants with emails.
45. Use tracking IDs
Most survey tools allow you to use tracking IDs, so that you can match an invitation with a completed survey. If you can use tracking IDs it’s a good idea to do so as this will allow you to track the effectiveness of invitations and means that you don’t have to recapture information that you might already know, such as someone’s email address.
46. Set a closing date
There is nothing like a deadline to induce action. Even if you don’t really have a closing date for a survey, it’s a good idea to set one so that it’s clear that would be participants must act soon.
If you like this article then please recommend and share it. You can find lots more articles like this on my blog: UX for the Masses