‎Refining Surveys using AI-powered Question-Level Suggestions

Refining Surveys using AI-powered Question-Level Suggestions

Updated

3 days ago

Question-Level Suggestions in Sprinklr Surveys help you identify and resolve potential issues within your survey. By analyzing the structure and content of your questions, they provide targeted feedback to improve clarity, relevance, and overall design. This ensures your surveys are optimized for better respondent engagement and higher-quality responses.

Question-Level Suggestions are especially valuable for improving feedback quality and minimizing drop-off rates. They provide a systematic way for Customer Experience professionals, product managers, and digital leads to refine survey design, enhance the respondent experience, and ultimately drive stronger business outcomes.

Note: Question-Level Suggestions currently provide feedback only on question-related issues.

Business Use Cases

Boost Completion Rates with Clearer Questions: Question-Level Suggestions enhance the clarity and tone of your surveys, which can greatly boost completion rates and yield more actionable feedback. By using Question-Level Suggestions, you receive specific recommendations to improve your questions. If your survey isn’t performing well, this tool assists you in pinpointing vague or poorly phrased items and allows for quick revisions, leading to a more seamless experience for respondents and improved insights for your team.
Streamline Beta Testing with Smarter Surveys: Question-Level Suggestions assist in enhancing your surveys by addressing problems such as repetitive text entry questions. During beta testing, obtaining clear and targeted feedback is essential. This contributes to a more seamless experience for participants and allows your team to collect more precise data, ultimately resulting in improved product decisions prior to launch.

Customer Experience and product teams can utilize Question-Level Suggestions to improve the quality of surveys and enhance the experience for respondents. This feature assists in identifying and addressing possible problems, such as ambiguous phrasing, repetitive questions, or inadequate structure, prior to the survey's distribution. By refining the flow and clarity, teams can secure more valuable feedback and dependable insights to aid in business decision-making.

Prerequisites

You will require Program Level View, Edit, and Delete permissions to gain access to it.:

View: Helps to view the feature.
Edit: Helps to edit the feature.
Delete: Helps to delete the feature.

Setting Up Question Level Suggestions

Hover over the Question-Level Suggestions icon in the bottom menu of any relevant question within the Survey Builder to see the number of flagged issues.
Click the Question-Level Suggestions icon to view any detected issues for the selected question.
Each detected issue is displayed as an Issue Card containing the following:
- Issue Description: Provides a detailed explanation of the identified issue.
- Suggestions List: Provides actionable recommendations to resolve issues.
- Navigation: Click on the side arrow button to navigate between different issues identified for a question.
Question-Level Suggestions track 5 different Issues:
- Moderate use of Text Entry Questions
- Choice Overload Avoidance
- Moderate Use of Matrix-Style Questions
- Clean and Concise Language
- Use Verbal labels for Rating Questions
- Sensitive Information Request

Note: No Question-Level Suggestions are shown for newly created surveys before any questions are added.

Question Level Suggestions

Survey Issue Heading	Issue Explanation	Issue Suggestions
Use verbal labels for rating questions	Respondents find it hard to infer the meaning of a rating scale if abstract numbers are used as labels. We find that, on average, completion rates start to drop, and respondents start being less deliberate in their responses when they feel that questions are ambiguous.	Consider using verbal labels in rating questions in order to eliminate ambiguity about scale meaning
Clear and Concise Language	Dynamically flags issues around spelling and grammar, tone, profanity, clarity of questions, unnecessary length of questions	Specific to questions
Moderate use of text entry questions	Respondents use a significant amount of mental energy when writing text. We find that, on average, completion rates start to drop, and respondents start writing far less in their responses when a survey has more than three open-text areas.	Consider deleting text boxes if they are not very necessary at a particular place . Consider using simple, straight forward, easy-to-answer question text to increase the likelihood of them being answered. Consider ensuring that the majority of the text entry questions are not response-required questions . Consider converting them into other question types if possible, reducing the cognitive load for the respondent. Example: Instead of asking “What is your age?” as a Text Entry question, a drop down can be asked, with different age categories the respondent might belong to)
Choice Overload Avoidance	Respondents use a significant amount of mental energy when deciding between multiple options. We find that, on average, completion rates start to drop, and respondents start being less deliberate in their responses when MCQ and Rank order questions have more than five answer choices.	Consider deleting answer choices if they are not very necessary. Consider combining two answer choices if they address very similar concerns. If all answer choices look equally important, consider breaking down the question into two smaller questions. Use simple, straight forward, easy-to-answer question text to increase the likelihood of them being answered
Sensitive Information Request	Some questions that may potentially be soliciting sensitive information (names, contact details, health, ethnicity, religious beliefs, political opinions, IP addresses, etc) have been detected. When designing surveys, it’s essential to adhere to GDPR guidelines to avoid collecting protected personal information unnecessarily.	Only ask for data that is strictly relevant to your objectives and avoid requesting information that could directly or indirectly identify individuals (such as names, contact details, or IP addresses) unless you have a clear and justified reason to do so. For any personal data collected, ensure you have obtained explicit, informed, and opt-in consent, providing participants with a clear explanation of what data is being collected, why it is needed, and how it will be used.
Moderate use of matrix style questions	Research has shown that response quality and completion rates both decline when questions use the matrix format. We’ve detected that the number of matrix rows contained in your survey may negatively impact the quality of data that you collect.	Consider simplifying possible responses (for example: Strongly Disagree, Disagree, Neutral, Agree, Strongly Agree could be simplified to Disagree, Neutral, Agree). Consider using separate multiple-choice questions where applicable, and try to keep matrix tables to a minimum.

Best Practices

Review and resolve questions with a higher number of question-level suggestions first to improve survey quality.
Use the Suggestions List to get actionable guidance on how to resolve specific issues and improve your survey questions.
Note: The suggestions provided are based on commonly accepted industry best practices. However, you should exercise your own judgment and validate them on your end before taking action.