A Discovery Survey helps you understand your users' needs and find patterns in what's important to them. This guide shows you how to analyze and sort the responses to identify top tasks and user challenges.
This is the goal of the analysis
Identify top tasks
Sort responses with Labels for statistics and filtering
Identify which categories dominate the answers
Find examples of the user tasks and patterns or subtasks within the user tasks
Understand user needs, target groups and challenges
Understand what tasks users want to solve.
Find explanations for why some users are unable to complete their tasks.
Use filters for target groups and devices to analyze differences
Present the answers in a meeting or workshop
Target top tasks
A natural next step is also to create a Top Tasks survey.
We recommend using a table view of the responses from one selected Discovery Survey.
Use Labels for categorization
Responses can be placed in a category by using Labels in Skyra. Labels will also give you statistics with the number of responses per label.
Create a new label
First you need to create the label using one of the following methods.
Select a row > Tag > Create new label > fill in
Press Label in the column > Create new label
Change labels in Settings > Labels
Use Label as a filter
Tap a Label to see all responses with it. In addition, you can select multiple labels at the same time with the drop-down menu in filter > label.
Example
Here, two labels are used to see those marked with both the user task transport and given feedback about a problem.
Tips for categorization
When you're analyzing responses from a Discovery Survey, we have these tips for working through the data.
Remove noise and personal information
Look for patterns
Sort answers into categories
Tips: Take notes as you analyze.
Continuously note which keywords were used to find categories, issues mentioned frequently, answers that can be used to illustrate important points, etc.
Start with broad categories
You may want to start with broad categories first and narrow down to specific categories afterwards. In the example below, we have started with Education as a category without specifying Labels for application, admission, exam etc.
Procedure for sorting
Remove noise
Identify and exclude irrelevant answers.
Use keywords to find noisy responses and give them their own label, such as "Irrelevant" or "Personal info".
Mask personal information in the answers to preserve the anonymity of the data.
Examples of search terms:
Numbers (0, 1, 2, 3, etc.)
Unusual letters (X, Z, W, Q, Y)
Email and personal identifiers (@, mvh, mob)
Look for patterns
Look for answers that are repeated often.
Compare with known tasks or hypotheses you have in advance.
Tips for effective searches:
Use parts of a word to include variations (e.g. "island" for Askøy, Skøyen).
Categorize place names, activities, or other relevant patterns with labels.
Sort the answers
Create broad categories first, such as "Hunting and fishing", and avoid small detailed categories such as "fishing license" or "hunting license".
Use labels to organize your answers. Examples of labels could be:
Outdoor: Hike, kayaking, biking
Place: Bergen, Oslo, Stavanger
Accommodation: Camping, hotel
Transport: Ferry, taxi, bus
Start with simple sorts and spend time interpreting complex answers later.
Mark issues
Identify responses that point to challenges or issues.
Give these responses specific labels to ensure they are followed up, such as "Technical challenge" or "Navigation issue".
Example of sorting
Name of Label: Outdoor
Tasks related to planning activities
Keywords to find matching answers
Hike, Hiking
Fish, Fishing
Kayaking
Biking
Swimming, beach
Result: All answers related to outdoor activities are given the label "Outdoor". You can then filter on this. In addition, we labeled "Example" to shorten the list to a representative sample to show in the meeting. This gives a clear picture of what users are interested in this category.
Name of Label: Education
Tasks related to information about education in a county. This is a broad category for all user needs related to education.
Result: We see that education is an important task and get a number of examples of what kind of information people are looking for.
How to deal with unclear or incomplete answers?
When analyzing a Discovery Survey, you're likely to encounter responses that seem unclear or incomplete. These can be challenging to categorize, but here's an effective method for dealing with them:
Use a temporary label: Start by sorting unclear answers into a separate category, for example by using a label like "Unsure". This makes it easier to clean up and focus on the more obvious answers first.
Analyze unclear answers later: After categorizing the clear answers, you can go back to the unclear ones. These often require more time and context to understand.
Look for patterns and context: Experience shows that unclear answers often become clearer when you look at several answers at once. For example, repeating words or phrases can provide clues:
An online retailer discovered that four-digit answers actually referred to postcodes that users searched for to find their nearest store. The task thus became "Store and opening hours".
This was revealed by looking at the URL where the answers came from, and some answers contained both four digits and a comment that they could not find any shops in their area.
One tourism operator received answers with only place names that at first seemed vague, but later they understood that the task was about "Things to see and do" at travel destinations or information to plan a specific activity, such as Trolltunga.
This was revealed by looking at the URL where answers came from and also a comment about what information they expected to find, but did not find.
Next steps: Top Task Survey
Once you have sorted and analyzed the responses from the Discovery Survey, you can move on to measuring top tasks with a Top Task Survey. In this survey, we allow the user to select their task from a list and provide feedback on it. Categorization and statistics are automatic and answers per user task can be read in the context of the user task in the Dashboard.
Read more about top tasks
Top task survey results