Thoughts on surveys

Surveys are great, if they make them concise and to the point.

What Makes a Great Survey

Keep it short, stupid

Your biggest concern is being clear and concise, or in finding the shortest way to ask a question without muddying its intent. It's not about reducing the character count; you must remove superfluous phrasing from your questions.

Survey length remains important for keeping abandon rates low. Think about the last time you sat around and excitedly answered a 10-minute survey.

Ask questions that fulfill your end goal

You need to be ruthless when it comes to cutting unnecessary questions from your surveys.

Ask this question about every question you include, "Does this question drive action once received?" If the answer is no, why are you asking it? The action doesn't have to get you calling the surveyee, but should drive trends to changes you want to make.

Construct smart, open-ended questions

Some of your most insightful feedback will come from open-ended questions. Nothing makes a survey more intimidating than a large text box connected to the first question. It's best to take on brief questions first to create a sense of progress, and then give survey takers who've made it to the closing questions the opportunity to elaborate on their thoughts.

One strategy is to get people to commit to a question with a simple introduction, and then follow up with an open-ended question such as, "Why do you feel this way?"

Ask one question at a time

We've all been hit with an extensive series of questions before: "How did you find on our site? Do you understand what our product does? Why or why not?"

It can begin to feel like you're being interrogated by someone who won't let you finish your sentences. If you want quality responses, you need to give people time to think through each individual question.

Bombarding them with lots of points to consider leads to half-hearted answers by respondents who will be looking to get through to the end --- if they even stay with the survey at all. Make it easy by sticking to one main point at a time.

Make rating scales consistent

Common scales used for surveys can become cumbersome and confusing when the context begins to change.

Here's an example I came across. While answering a survey's initial questions, I was told to respond by choosing between 1-5, with 1 = "Strongly Disagree" and 5 = "Strongly Agree."

Later on in the survey I was asked to evaluate the importance of certain items. The problem: Now 1 was assigned as "Most Important," but I had been using 5 as the agreeable answer to every previous question.

This is incredibly confusing.

Avoid leading and loaded questions

Questions that lead respondents towards a certain answer due bias in their phrasing are not useful for your surveys. SurveyMonkey offers a great example of a leading question to avoid: > We have recently upgraded SurveyMonkey's features to become a first-class tool. What are your thoughts on the new site?

Remember to cut out language that caters to ego or contorts a respondent's understanding of what's being asked. To avoid loaded questions, stay away from any presupposed facts or assumptions.

A well-known example on disciplinary action with children is as follows: > Should a smack as part of good parental correction be a criminal offence in New Zealand?

The assumption here is that smacking a child is inherently a part of "good" parental correction, when in fact that is the opinion being argued. You can avoid loaded questions in your surveys by eliminating emotionally charged language that hints at preferences or assumed facts.

Make use of Yes/No questions

When you are asking a question that has a simple outcome, try to frame the question as a Yes/No option.

The Survey Monkey study showed that these closed-ended questions make for great starter questions because they are typically easier to evaluate and complete.

These questions can also be used to qualify the respondent with less of an ego bias, such as asking a question like, "Are you considered an expert in [blank]?" vs. "What level of expertise do you have in [blank]?"

Get specific and avoid assumptions

When you create questions that implicitly assume a customer is knowledgeable about something, you're probably going to run into problems (unless you are surveying a targeted subset of people).

One big culprit is the language and terminology you use in questions, which is why I'd recommend staying away from industry acronyms, jargon, or references.

One of the worst assumptions you can make is to assume people will answer with specific examples, or to explain their reasoning. It's better to ask them to be specific, and let them know you welcome this sort of feedback:

How do you feel about [blank]? Feel free to get specific; we love detailed feedback!

Timing is important

The Survey Monkey study found the highest survey open and click-through rates occurred on Monday, Friday and Sunday respectively.

There was no discernible difference between the response quality gathered on weekdays versus weekends, either, your best bet is to seek out survey-takers first thing during a new week or to wait for the weekend. Perhaps Monday has such high response rates because nobody feels like working.

Give them a bonus

It sometimes makes sense to entice customers to take your survey: a variety of data show that incentives can increase survey response rates by 5 to 20 percent. These incentives could be a discount, a giveaway or account credit.

A valid fear is that a freebie may detract from the quality of responses, but a few studies show that this isn't likely to be the case. Make incentives something you can financially handle, such as an extended trial of your software for a period of time.

Ways of Surveying

Three ways of surveying your end users are:

• Immediate feedback on any page
• After some event like the close of an incident or request
• On a frequency survey

Survey's can be helpful and I'd suggest having at least one of the below set up to get a indicator of where you are, and how you can improve.

ServiceNow's built in survey engine has some room for improvement but is good enough for most survey's needs. The problem is they are not nearly as intuitive to set up as SurveyMonkey.

Based on the feedback that you begin to receive you can decide if extra questions should be asked. However, you want to make sure that Surveys are simple and short.

Follow up on any comments that show someone had less then a stellar experience to get more details Adjust your question where you find a pattern with comments that people are leaving to ensure you are capturing the right data

Immediate Feedback

I love the idea of having a simple "How can we make this page better for you?" as pictured below. I'd have it have a thumbs up, and thumbs down to submit, and on press of enter submit w/out thumb up or down value.

After an Event

What kinds of events?

• Resolution of an Incident accepted.
• Close of an Incident.
• Close of a Requested Item.
• Close of a Project.

The key here is asking right away to keep information fresh in the users mind.

Regularly Scheduled

These are great in keeping a heart beat on the satisfacation and other metrics gathered from the survey.

Pretty much per application, I'd ask some questions monthly or quarterly.

Examples

List of questions seen

Most Common Survey

Survey used after the close of an incident

Survey used by Servicenow internally

Wordpress Minneapolis 2017

This survey isn't short, and asks more than one question at a time. It also asks questions regardless of answers. Don't do this.

Upgrade Instance Survey

This survey is concise and satisfying.

I know at a past place, I set up a business rule on a question to notify the manager of the incident that spawned the survey to followup. That was a easy big win

Further Reading

• The 5 Best Ways to Get Feedback from Your Customers
• Survey Question Examples
• CSAT Question and Answer Option Menu
• ServiceNow Survey Docs
• Wordpress Survey