Multiple-Choice Survey

The method is available in two verticals: in General Research you open it with the Survey card, and in Fashion with the Multiple Choice card. The question editor is the same in both; the differences are in the credit calculation and the report. The Fashion report has different breakdown dimensions and has no Statistics or Insights tabs and no low-response-rate warnings.

When to use it

• When you want to see the response distribution for questions such as purchase channel, frequency of use or brand preference.
• When you want to ask rating questions such as agreement, likelihood to recommend or satisfaction scales.
• When you want to show packaging, a product photo or a campaign visual and measure reactions.
• When you want to see how responses differ by age, socio-economic status (SES) or an attribute you define yourself.

For price questions, Price Analysis (WTP) is a better fit, and for ranking features by importance, Priority Ranking (MaxDiff). To compare methods, see Choosing the right method.

Before you start

• Population: You can select an existing population or create a new one during setup. In Fashion, populations are archetype-based. For details, see Creating a population.
• Custom feature: If you want to split the results by your own hypothesis (e.g. “Prefers eco-friendly products”), add this feature when you create the population in General Research. For details, see Custom features.
• Images: Prepare the images you want to show with your questions in JPG, PNG or WebP format, each up to 5 MB.
• Credits: Make sure your balance covers the estimated credits.

You can also have Argus set up the study for you in a conversation. To run a saved survey without going through the wizard, see Quick Research.

Setting up the study

In General Research

1. In the sidebar, click New Research.
2. Select the General Research card.
3. On the Choose Your Research Type screen, click the Survey card.
4. On the Choose Population Mode screen, choose Select Existing Population or Create New Population.
5. On the design screen titled Multiple-Choice Survey, write your questions (see Writing questions below).
6. Click Next Step.
7. In the dialog that opens, click Continue.
8. On the Research Preview screen, check the Person Count, Question Count and Credit Usage values in the Experiment Summary card.
9. Click Start, then Continue.

In Fashion

1. In the sidebar, click New Research.
2. Select the Fashion card.
3. Click the Multiple Choice card.
4. On the Choose a population screen, choose Use existing population or Create a new population.
5. On the Anket Tasarla (Design Survey) screen, write your questions.
6. Click Next Step.
7. On the Research Preview screen, click Start.
8. Confirm the start in the dialog that opens.

Writing questions

1. Type the survey’s name in the Survey Title field. The Questions: counter shows how many questions you have added.
2. In the Create New Question card, type your question in the Question Text field.
3. To add an image, drag the file onto the Question photos (optional) area, or click the area and select the file.
4. Type an option in the option box, then press Enter or click Add Option. Add at least 2 options.
5. For Answer Type, choose Single Select (the default) or Multi Select.
6. Click Add Question to Survey.

Questions are numbered Q1, Q2 and so on in the Survey Questions list. On each row, you can change the answer type, edit the question with the pencil icon or delete it with the bin icon. There is no separate rating (Likert) question type; a scale question is a Single Select question whose options are the scale statements.

Importing from a saved survey

Every multiple-choice survey you write in a study is saved to the Surveys library. To reuse a survey, choose it from the Select a survey… list under Import from Saved Survey and click Import. Its questions are copied, together with their images and conditions, to the end of the current list. For the library, see Survey library and question design.

AI suggestions

• Suggest Question with AI: Appears once you have added at least one question. Fills in the form with a follow-up question, up to 5 options and an answer type.
• Suggest Options with AI: You need to write the question text first. Adds up to 5 new options without repeating the existing ones.
• AI Instructions: In this box, which you open with the gear icon, you can write guidance (e.g. “Focus on price sensitivity”). Both suggestions take this guidance into account.

Suggestions only fill in the form; the question is not added to the survey until you click Add Question to Survey.

Conditional questions

A condition shows a question only to personas who chose particular options in an earlier question.

1. On the question’s row, click the Koşul (Condition) button.
2. From the Soru seçin (Select a question) list, choose an earlier question.
3. Under Şu seçeneklerden biri seçilirse göster (Show if any of these options is selected), tick one or more options.
4. Click Tamam (OK). A “Koşullu: Q1 → …” (Conditional: Q1 → …) badge appears in the list.

To remove the condition, click Koşulu kaldır (Remove condition) in the same panel.

If a persona chose any of the options you ticked, the question is asked and the persona is reminded of their earlier answer. Conditions can form a chain: if a question is not shown to a persona, the questions that depend on it are not shown either. When you try to delete a question that other questions depend on, the warning “Bu soruya bağlı koşullar da silinecek. Devam edilsin mi?” (Conditions linked to this question will also be deleted. Continue?) appears.

Settings and limits

Credit calculation

• In General Research, conditional questions count towards the number of questions, even though some personas do not see them.
• The Credit Usage value on the Research Preview screen takes the form “balance/estimated credits” (e.g. 1.200/1.000). If your balance is not enough, a red icon appears next to the value; hover over it to see the Buy Credits button.
• In Fashion, the Start button is disabled if your balance is not enough. In General Research, the button stays clickable, but if your balance does not cover the estimate, the study fails and no credits are deducted.
• Credits are deducted after the responses are saved. In Fashion, no credits are deducted for a failed study; in General Research, a study that fails before its responses are saved does not spend credits either. In General Research, if the study fails at the analysis step after the responses have been saved, the deducted credits are not refunded automatically.

For your balance and buying credits, see Credits.

Reading the results

While the study is running, the progress screen (Experiment Processing) shows how many personas have been processed; if you leave the screen, the study carries on in the background. When it finishes, a Go to Reports button appears. You can find the report later on the Reports page, on the card with the Survey MCQ badge. If the study fails, the Experiment Failed screen opens: in Fashion, click Retry; in General Research, click Back to Preview, then Start again. In General Research, if the study failed at the analysis step after the responses were saved, submit a request in the Billing category on the Support page before you restart it; restarting can generate the responses again from scratch and deduct the credits again.

The report header shows your survey’s name. At the top are these summary boxes:

If fewer than 90% of the personas who saw a question answered it, the question is listed in the Low Response Rate Warnings box with its response rate and number of missing responses. The box shows at most 3 questions.

Result tabs:

• Response Distribution: Each question sits in a card that you can expand and collapse. The card shows the number of responses, the response rate, the number of missing responses, the question images, and each option’s count and percentage. For conditional questions, the card also states who the question was asked to. Use Expand All and Collapse All to expand or collapse every card at once.
• Insights: Appears in General Research reports when there are findings. It is made up of Majority Response (the most chosen option has a share of 50% or more), Missing Data (at least 20% of the personas who saw the question did not answer), High Diversity (at least 3 options were chosen and the most chosen option’s share is below 35%) and Demographic Difference cards. Demographic Difference shows cases where an option’s share in a segment is at least 40% and differs from the overall share by at least 10 percentage points. This finding counts as reliable only when the p-value is below 0.05 and the segment has at least 30 personas. If there are no reliable findings, the strongest candidates are listed under the Limited Data Warning heading.
• Data Table: All results appear in a single table with Question, Option, Count and Percentage columns. Use the Filter… box to search the table.

The results are responses generated by language models. Use them to see the direction and size of trends. For details, see Accuracy and limitations.

Exploring by segment

• Demographic Breakdown: Choose a dimension from the Select Demographic list; for each question, the segments’ response distributions are shown as side-by-side bars. In General Research, the dimensions are age, gender, education, income group, SES, market segment and custom feature. In Fashion, they are archetype, channel, style confidence, income, age, gender, education and SES.
• Crosstab: Make a selection in the Question and Demographic fields; a heatmap and a stacked bar chart appear for the question and dimension you chose.
• Statistics: Appears in General Research reports when there are test results; the number in the tab name is the number of significant relationships. For each question–demographic pair, it lists the chi-square test result (χ², p-value, Cramér’s V) and the Effect Size (negligible, weak, moderate, strong). A relationship is considered significant if the p-value is below 0.05. Click a row to open the Demographic Breakdown view for that dimension. Multi Select questions are not included in this test.

For details on breakdown dimensions, see Segmentation.

Exporting

• Download PDF: If suitable segment dimensions are available, the PDF segment selection dialog opens; tick the dimensions you want to include in the report and click Generate PDF. The PDF report contains the response distribution, the breakdowns you selected, the key findings and a summary.
• Download CSV: Downloads the table on the Data Table tab.
• Download Excel: Prepares an Excel file with six sheets; the sheet names are in English.

In General Research studies, the Population and Raw sheets have a Custom Feature column. This column shows the persona’s custom feature and is empty for personas without one. Because the column is filled in when you download the file, it also appears for your older studies. On the Raw sheet, Multi Select answers are separated by semicolons (;). A “—” means the question was not shown to that persona because of a condition, and an empty cell means the answer is missing.

Tips and common mistakes

• Ask one thing per question. Instead of “How are the price and quality?”, write two separate questions. Also make sure the options do not overlap.
• Add a “None” or “Undecided” option. Personas are told that these options are entirely valid answers too.
• Use standard wording in scale questions. Write between 3 and 7 options and prefer well-known scale wording such as “Strongly agree … Strongly disagree”. The system recognises these questions as scales and asks them in a way that makes personas less likely to bunch on the middle option.
• Use Multi Select only when you need it. These questions are not included in the test on the Statistics tab.
• Don’t rely on option order. The order of the options is shuffled for each persona. In General Research surveys without conditions, the order of the questions is shuffled too. Do not write position-dependent options such as “All of the above”, and make each question understandable on its own.
• Keep the text of options used in conditions unchanged. Conditions match options by their text. If you change or delete an option of an earlier question that a condition uses, that option is removed from the condition; if no options are left in the condition, the condition is removed.
• Plan enough personas for segment findings. Differences in segments with fewer than 30 personas are not considered reliable. Because many question–dimension pairs are tested at the same time, a few “significant” results can appear by chance; prioritise those with a moderate or strong effect size.
• Cut unnecessary questions in General Research. Credits grow with the number of questions; trim your question list before you make the population bigger.

Related pages

• Survey library and question design
• Quick Research
• Segmentation
• Exporting and downloading
• Credits