Inline Survey Editor manual

Purpose

The Inline Survey Editor creates questionnaires whose questions appear as highlighted phrases inside a short story. Respondents read naturally, open each phrase, save an answer and can revise it before submitting. A survey can be shared as a hosted link, downloaded as one standalone HTML file, or both.

Nothing is published while you edit a local draft. The live preview is for continuous checking; Full preview opens the exact respondent experience.

First steps

On a new browser profile, First steps opens automatically. Completing it opens the New survey dialog. First steps and this complete Manual remain available from the editor header.

1. Start with Blank survey, copy a built-in template, or describe the survey to the AI assistant. Community event is a useful learning template because it demonstrates several question types, including a multiple-answer question.
2. Give the draft a recognizable local name. A template is copied once, so all of its text, questions and presentation can be changed safely.
3. Write the introduction and story paragraphs. Insert each question marker at the exact phrase respondents should select.
4. Choose an appropriate type for every question. Use one answer for mutually exclusive options and enable Allow multiple answers when several options may genuinely apply.
5. Set the respondent language, layout, appearance and accent. Decide whether to show an approximate duration and a survey print button.
6. Resolve every Validation item. Test the live preview and Full preview on desktop and mobile. Answer every question, select several options in a multiple-answer question, skip optional questions, edit answers, reload, reset and submit.
7. Choose Publish & deliver. Select Database, Email or Database + email based on whether dashboard responses and exports must remain available.
8. Enter the management email and send the activation message. Open its link; the still-open wizard may need a few seconds to show the confirmed state.
9. Open or configure the hosted survey, download the standalone HTML if needed, and store dashboard access safely. Keep a downloaded JSON backup of important editable drafts because browser-local data can be cleared or lost.

AI survey assistant

The AI assistant can create a separate draft from an ordinary description or one public webpage. State the subject, intended respondents, desired language and what you want to learn. Never include confidential or personal information in the request.

The assistant can also explain how to place a finished survey on a website. Ask, for example, "How do I embed this survey on my website?" It will explain the appropriate hosted-link or iframe steps without changing the survey. Survey design remains its primary job: if a request combines creation and embedding, it first produces the complete survey proposal; ask for embedding help after activation. The public Embedding a survey guide begins with simple sharing and continues with iframe, HTML and responsive-layout examples.

Useful creation prompts include:

• Create a short French visitor survey about arrival, facilities used, staff helpfulness and one optional suggestion.
• Create an English event survey. Use multiple answers where respondents may have attended several activities.
• Create a German product survey with a dark appearance and no personal-data questions.

Use Create new for an independent survey. Use Edit current for focused changes such as adding a question, simplifying wording, changing suitable question types, adjusting the design or translating the open draft. Review the complete proposal before applying it; until then the editor is unchanged. One Undo can reverse an applied assistant edit.

The assistant decides between one and multiple answers from the meaning of the question. It should use multiple answers only when several options can be true at the same time, such as activities attended, features used or reasons for a choice. Satisfaction levels, age ranges, yes/no questions and primary choices normally require one answer. You can state the desired mode explicitly whenever it is important.

If a prompt appears to target the wrong mode or a different selected draft, the assistant asks before continuing. If the draft changed while a proposal was waiting, choose whether to update the proposal, apply the older proposal or cancel. Conversations and applied history stay in this browser and never enter the generated survey or downloaded JSON.

Editing the story and questions

The story is made from ordinary paragraphs. A marker such as [[facilities]] connects one highlighted phrase to the question whose ID is facilities. Use the question selector below a paragraph to insert a marker at the text cursor. If a paragraph contains exactly one valid marker, the selector follows it and Edit question jumps directly to that question.

Every question needs one unique ID and exactly one matching story marker. Let the editor create IDs whenever possible. Renaming an ID through the editor also updates its marker. Removing a question while its marker remains creates a Validation error; remove or replace that marker as well.

Use the arrow controls to reorder paragraphs or questions. Reordering questions does not change their markers. Keep clickable phrases neutral and natural in the surrounding sentence: the phrase should identify the subject without implying that the respondent liked, used or experienced it.

Editor fields with an × clear only that field. Remove deletes a complete paragraph or question after confirmation. Download JSON before extensive edits when you want a portable recovery point.

Choosing question types

Use differentiated answers when labels carry useful meaning. A satisfaction progression, a yes/no decision, named categories and several applicable options all belong to Choices, with the selection mode chosen separately.

Use a numeric scale when evenly spaced positions are meaningful. Buttons make all positions visible and work especially well for three or five points. A discrete slider is useful for gradual seven-to-fifteen-position ratings. Give both endpoints precise labels and keep related scales in a consistent direction.

Use an input field for short text, long text, numbers, dates, times or email addresses. Configure meaningful limits, increments, units and help text. Long suggestions and personal information should normally be optional.

Use a checkbox for one explicit Boolean confirmation, such as acknowledging information or requesting occasional updates. A checkbox is not a list of several answer options; use Choices with Allow multiple answers for that.

A required question must be completed before submission. An optional question can be skipped explicitly. If some respondents may not have used a service, either make the question optional or include a clear Not used / Not applicable answer. Do not force them to rate an experience they did not have.

Single and multiple answers

Choices use one answer by default. Enable Allow multiple answers only when the respondent may truthfully select several options. Good examples are activities attended, services used, desired features or several reasons for a decision. Questions such as overall satisfaction, age group or principal transport mode should normally remain one-answer.

In a multiple-answer question the respondent toggles options and then chooses Save answer. A required question needs at least one selection. An optional question can be skipped. Reopening the question shows the saved selections; Clear answer removes them all.

Keep options independent and avoid contradictory combinations. If None of the above or Not applicable must exclude every other choice, word the question carefully and verify the respondent experience in Full preview; the current selection control does not impose an exclusive option automatically.

The dashboard and response email show every selected label. CSV and TSV join the labels with " | "; JSON retains an array. Statistics count each selected option. Consequently, option percentages in a multiple-answer question can add up to more than 100 percent.

Older surveys remain one-answer unless multiple answers are explicitly enabled. In advanced JSON editing, a Choices question uses "multiple": true or false; leaving the property absent in an older definition preserves one-answer behaviour.

Respondent languages

The respondent runtime supports English, German, French, Spanish, Italian, Portuguese, Dutch, Swedish, Danish, Norwegian, Polish, Czech, Finnish, Romanian, Hungarian, Greek, Turkish and Russian. The editor interface is available in the first six languages.

Survey language controls fixed respondent buttons, validation, accessibility messages and localized date/time behaviour. Authored titles, paragraphs, questions and answers are not automatically translated when the setting alone changes. Translate the content as well, manually or with Edit current.

Content may be written in another language, but one supported fallback must be selected for fixed controls. Test that combination carefully and never describe the unsupported content language as a fully localized runtime language.

Drafts, backups and published versions

Drafts, local names and editing history are stored in the current browser. Use Download JSON as the portable editable backup, especially before clearing browser data, changing device or making extensive edits.

Activation locks the exact published definition. Later work must not silently alter what respondents are already using:

• Open in editor loads an active, paused or closed survey for inspection. The first content change automatically creates a separate local new-version draft.
• Create editable copy creates that independent draft immediately.
• Rename changes only the local management label, not the public survey or its submitted data.
• Duplicate creates a separate starting point and deliberately omits assistant conversation history.

Active surveys accept responses. Paused surveys temporarily refuse online responses and can be reactivated. Closed surveys permanently refuse new online responses. A downloaded standalone file remains independent, although online submission still follows the published survey state.

Design and editor settings

Choose a layout that suits the subject and audience, then test both light and dark appearances when Automatic is used. The accent should remain readable against buttons, highlighted phrases and focus indicators. Text size changes the respondent presentation; editor appearance changes only the editing screen.

Settings are browser-local. They control interface language, Auto/Light/Dark editor appearance, editor accent, navigator side, tooltips, motion preference, reopening the last survey, preview defaults and preview geometry. Resetting interface preferences or clearing local drafts does not delete published surveys or server responses, but local drafts without a JSON backup are lost.

The live preview can use Fit panel or a fixed zoom. Fit panel resets both scroll directions and fits the complete survey width. With fixed zoom, use the compact scrollbars or drag an empty preview area to pan. The wheel scrolls vertically, Shift + wheel horizontally and Ctrl/Cmd + wheel changes preview zoom.

Validation and testing

Validation must show no errors before publishing. It checks missing or duplicate markers, question structure, answer lists, numeric ranges, labels, required settings and other conditions needed by the generated survey.

Use this final test checklist:

1. Read every paragraph and open every highlighted phrase.
2. Test one-answer and multiple-answer questions, including Save answer and Clear answer.
3. Verify required questions block submission and optional questions can be skipped.
4. Reopen and change saved answers.
5. Check number, date, time and email limits where used.
6. Reload with partial answers and confirm that answers remain.
7. Test Reset answers.
8. Check desktop and mobile widths, keyboard navigation, focus visibility, contrast and reduced motion.
9. Use Full preview for the final interaction and print test.
10. Submit one complete test response after activation and verify the expected email or dashboard result.

The optional duration is an estimate based on reading length and question type, not a promise. The optional respondent print button uses a clean layout and includes answers already entered in that browser, including multi-line text.

Publishing and delivery

Choose delivery before activation according to how results should be handled:

• Database retains answers for the private dashboard, statistics and exports.
• Email sends each response by email but does not retain answer content for dashboard statistics or exports after delivery processing.
• Database + email retains dashboard data and also sends the response by email.

Publish & deliver reviews the survey, delivery, management email and activation. Activation locks that version. Keep activation and access messages private. After opening the activation link, allow the original wizard a few seconds to refresh its status before retrying.

Hosting creates a stable public respondent link. Download standalone HTML creates one portable file that also works offline. You may use either or both. Do not edit the generated HTML manually; return to the editor and publish a new version when content must change.

Availability, access and sharing

Public survey settings can define an opening time, closing time, readable public address and access code. Empty opening means immediately; empty closing means no expiry. Dates use the selected time zone. Online availability is enforced by the service; an offline file can only use the device clock as guidance.

One access code can protect the hosted survey, future standalone downloads or both. Hosted protection is enforced online. Protection embedded in standalone HTML is only a basic offline deterrent and can be bypassed by someone who edits the file.

Share provides the respondent link, native sharing when available, email, a public QR code, iframe code, a text-link snippet and a button snippet. The public QR never grants dashboard access. Removing hosting hides the hosted page but does not close the survey or invalidate already downloaded standalone files.

Dashboard and access

The dashboard shows survey state, delivery status, statistics, retained responses, exports and lifecycle controls. Expand an individual response to inspect its accepted answers. Filters can limit responses by survey language or search matching questions and answers.

The dashboard address identifies the survey but does not itself provide a valid browser credential. A temporary access link authorizes one browser session and must be kept private. Lock on this device removes only the current browser's credential. Lock on all devices revokes every current dashboard session.

Open in editor is useful for inspecting the published definition without immediately creating another draft. Create editable copy is preferable when you already know that a revised version is needed.

Results, statistics and exports

Statistics separate answered and skipped responses. One-answer distributions allocate each respondent to one option. Multiple-answer distributions count every selected option, so their percentages can exceed 100 percent in total. Always interpret the question wording and response count together.

CSV/TSV export offers semicolon, comma and tab separators. Choose the separator that imports correctly in your spreadsheet application. CSV and TSV contain readable labels; multiple selections are joined with " | ". JSON retains the full response and question metadata and preserves multiple selections as arrays.

Exports are available only when the selected delivery mode retains answers and at least one accepted response exists. Printing the dashboard produces a simplified results and statistics layout.

Privacy and safe use

Collect only information needed for the stated survey purpose. Make personal or sensitive questions optional unless they are genuinely essential, and explain why the information is requested. Do not put confidential or personal data into AI assistant prompts.

Keep activation messages, temporary dashboard access links and downloaded response exports private. A public survey link or respondent QR code can be shared freely; it does not grant dashboard access.

Generated surveys validate their structure and submitted answers. Avoid manual changes to downloaded HTML because they can damage validation or submission. Use the editor and publish a separate version instead.

User troubleshooting

If something does not work:

1. Read and keep the exact visible message.
2. Reload once and try the same action again.
3. Confirm that Validation has no errors and use Full preview.
4. Check whether the survey is active, paused, closed or outside its availability window.
5. For email activation, wait a few seconds for the open wizard to refresh.
6. For missing dashboard statistics or exports, confirm that delivery retains answers and that at least one response was accepted.
7. If support is needed, provide the action, visible message, approximate time, browser and whether the problem occurs in live preview, Full preview, hosted survey or standalone HTML. Never send private access links or respondent answers in an ordinary support message.

Older supported survey files continue to open with their original one-answer behaviour. Keep a JSON backup before importing or extensively revising an older survey.