Survey Flow Editor

1. What Is the Survey Flow Editor?

The Survey Flow Editor (top navigation Survey) is the central workspace where you build and organize your survey. It shows every item in the order participants will encounter it, grouped into pages. Each item, whether a question, a text page or a choice task, appears as a card that you can reorder, edit, deactivate or delete.

Buttons on a page header

Buttons on an item card

The two logic links appear only on data-collecting questions. Text & Media, choice tasks and the End Page get Save as Template and Duplicate only. Choice-task cards also have a Show/Hide Tasks chevron. When Between-Subject Conditions are defined, the card footer additionally shows condition pills (All, A, B ...) that restrict the item to certain conditions; a click saves at once. Ctrl+click (Cmd+click on a Mac) on a card or on an entry in the Add Item panel opens it in a new tab. At the end of the flow, Add Page inserts an empty page.

Key Concepts

2. Adding Items to Your Survey

The Add Item panel sits on the right of the editor. On screens narrower than about 990 pixels it is hidden; open it with the Add & Manage button . It has two collapsible sections, Survey Questions and Conjoint. The search box (Filter items...) filters the list as you type and also knows method names, so typing "Kano", "PSM" or "Regex" finds the right type. The panel header links to this guide and to Templates , your saved items.

There are two ways to add an item.

Option A: Drag it into the flow

Press and hold an item type in the panel, then drag it into the flow. Drop it exactly where you want it: between two items, or onto a page boundary to start a new page. Choicery creates the item at that position with default settings and opens its editor. Constant Sum, Market Research and End Page open their add form first; a choice task opens as a draft and only enters the flow once you save it.

Option B: Click the item type

Click an item type to open its add form. On Save, the new item is placed just before the End Page: on the last page if that page is still empty (for example the empty first page of a new study), otherwise on a fresh page. You can then drag it wherever you like. Evaluation opens its own full-page editor, and the conjoint types open their configuration screen. New Page / Page Break opens no editor at all; it inserts the page break immediately.

What every item editor has

All standard item editors share the same frame: Question Name * at the top, the type-specific fields in the middle, a collapsible Footer for small print shown below the question, and at the bottom Cancel, Preview (shows the unsaved state) and Save . Large text fields (Question / Instruction) have a full formatting toolbar: Bold, Italic, Underline, Strikethrough, Text Color, Highlight, Font Size, alignment, bullet and numbered lists, Insert Link, Insert Image and Clear Formatting.

Lists of answer options, statements or items all work the same way. Each entry is one row with these controls:

3. Survey Structure: Pages & Navigation

A survey is divided into pages. Each page is one screen. Items between two page breaks (or between the start or end of the survey and a page break) form one page.

How Pages Work

• All items on a page appear at once. The participant answers them, then clicks Next.
• Page breaks create the boundary between pages. Without them, every item stacks onto one long page.
• Conjoint tasks (CBC, MaxDiff, and so on) and Market Research items always occupy their own page automatically.
• An End Page is a final screen, usually a thank-you message. A survey can have several End Pages with different outcomes (section 18).

A Sensible Structure

The Pizza study is ordered so the engaging, attention-hungry parts come first and the background questions come last:

Page Timers

The clock button in a page header opens Page Timer, which sets a minimum wait time before participants may continue: tick Enable timer, set Wait time (seconds) between 1 and 600 (default 30), and decide whether to Show countdown to participant (ticked by default). For choice blocks, the wait applies to every task screen, including the follow-up screens in two-block mode and the Anchored MaxDiff calibration screen.

4. Text & Media

The Text & Media item displays content without collecting a response. Use it for introductions, instructions, or consent text. It holds rich text (images go inside the text via Insert Image) and optional media: an uploaded video or audio file, or a YouTube/Vimeo embed. The Pizza study opens with a Text & Media page that welcomes respondents and embeds a short YouTube video.

Field / button	What it does
Question / Instruction *	The text of the page (required, up to 50,000 characters, at most 10 images).
Add Media	Opens Insert Media with three tabs: Video Upload (MP4, WebM, MOV, up to 50 MB), Audio Upload (MP3, WAV, OGG, M4A, up to 25 MB) and Video Embed (YouTube incl. Shorts, or Vimeo, with a Preview button). Confirm with Insert Media.
Remove	On the media card: removes the media again.
Position	Before Text or After Text (default).
Playback Settings	Uploaded videos only: Autoplay, Loop, Muted, Show Controls (on). Browsers only autoplay muted videos; the editor reminds you.
Width	Slider from 25 to 100 percent (default 100).

A project can hold up to 100 media files and 500 MB of media in total.

Storage: none. This item is display-only.

5. Single Choice

Single Choice presents a list of mutually exclusive options; the participant picks exactly one. It renders as radio buttons by default, or as a dropdown for long lists. The Pizza study uses radio buttons for diet, pizza_habits, gender and children, and a dropdown for pizza_order (How often do you order pizza?).

No.	Button	What it does
1	Import from TXT	Replaces all options with the lines of a text file.
2	T Formatting	Formats this option or inserts an image into it.
3	Add specify text input	This is the "Other, please specify" button. Click it on the option that needs a follow-up field. The button turns blue and the settings row 5 appears below the option. Click it again to remove the field.
4	Remove option	Deletes the option (greyed out here because a Single Choice needs at least 2 options).
5	Specify settings	Type: Text or Number. Placeholder: grey hint inside the field (default "Please specify", up to 200 characters). Width: 40 to 1000 px, empty means auto.
6	Add Option	Adds an empty option at the end.

Key Options

6. Multiple Choice

Multiple Choice lets participants tick one or more options. It renders as checkboxes. Use it when answers are not mutually exclusive. The Pizza Pizza_toppings item asks which toppings belong on a pizza, in two columns, with an "Other" option whose specify field opens when the box is ticked.

No.	Button	What it does
1	Import from TXT	Replaces all options with the lines of a text file.
2	T Formatting	Formats this option or inserts an image into it.
3	Add specify text input	Makes this option an "Other, please specify" option. In the Pizza item it is switched on for option 12 "Other" (blue button). Write the option text yourself, for example "Other" or "Something else".
4	Remove option	Deletes the option.
5	Specify settings	Type (Text or Number), Placeholder and Width of the follow-up field.
6	Add Option	Adds an empty option at the end.

Key Options

7. Grid / Likert Scale

A Grid (Likert) item presents several statements as rows and a shared scale as columns. Participants rate every row on the same scale, which is ideal for agreement, satisfaction, or importance batteries. The Pizza food_involvement grid rates eleven statements on a 7-point Strongly Disagree to Strongly Agree scale.

No.	Button	What it does
1	Import statements	Replaces the statements with the lines of a text file.
2	T Formatting	Formats this statement.
3	Remove row	Deletes the statement.
4	Add Row	Adds a statement at the end.
5	Import answer options	Fills the scale labels from a text file with 2 to 10 lines and sets Scale to the number of lines.

Key Options

The collapsible Layout Options group controls the visual detail. Text and spacing (Font Size, Row Padding) applies everywhere. Table (larger screens) holds Items Column Width (%) (10 to 80, default 45), Column Headers Alignment, Response Cell Width and Sticky Header for long grids; it only affects the table view.

8. Bipolar Scale

A Bipolar Scale (semantic differential) places opposing words at each end of a scale and asks participants where their impression falls. The Pizza Pizza_experience item rates the last delivery on eight pairs such as Cheap / Expensive and Slow / Fast, using a 7-point slider.

Button	What it does
Item card	Pairs are grouped into items. Each card has an "Item N" badge and an item text, for example the instruction above the pairs.
Left / Right	The two opposing words of a pair ("Pair N"). Both have a T Formatting button.
Remove pair	Deletes the pair (shown when an item has 2 or more pairs).
Add Pair	Inside the card: adds a pair to this item.
Add Item	Below all cards: adds another item group.
Remove item	Deletes a whole item card (shown when there are 2 or more items).
Input Type	Radio Buttons or Slider, followed by Min and Max.

Key Options

9. Ranking

A Ranking item asks participants to put a set of items in order of preference or importance. The Pizza pizza_order_ranking item lists ten aspects of ordering pizza and asks for the top 5 (Ranking Mode Top N, Interaction Style Drag & Drop).

What participants see

With Drag & Drop, the computer view has two panels. On the left is the pool of all items, each with a grip handle. On the right are the numbered ranking slots between the top label ("Most important") and the bottom label ("Least important"). Participants drag an item from the pool into a slot; a placed item gets its rank number and an button that sends it back to the pool. Items inside the ranking can be dragged to change their order. With Top N there are exactly N slots, so the Pizza item shows 5 slots for 10 items.

On phones, tablets and narrow windows the same item turns into a tap list (Figure 16). The ranking comes first, headed Your Top 5 with a counter ("2 of 5 selected"), followed by Available items. Participants tap + to add an item to the next free slot, move it by its handle and remove it with ×. A participant who tries to continue too early sees "Please rank exactly 5 item(s) before continuing."

The other Interaction Styles look the same on every device: Reorder shows one list of all items that participants sort (Full Ranking only), and Dropdown gives each item a rank dropdown.

No.	Button	What it does
1	Import from TXT	Replaces the items with the lines of a text file.
2	T Formatting	Formats this item.
3	Remove	Deletes the item.
4	Add Item	Adds an item at the end.
	Interaction Style	Drag & Drop, Reorder or Dropdown. The line below explains the chosen style; the line says how it works on phones.
	Ranking Mode	Full, Partial or Top N, with the number field next to it.

Key Options

10. Text Input & Formats

Text Input collects free-text answers. A toggle at the top of the editor switches between Single Line (names, short phrases) and Multi Line (comments, open-ended feedback). The Pizza favorite_topping item uses a single line with the lead-in text "My favorite pizza topping is/are".

Field	What it does
Single Line / Multi Line	Switches the field type.
Placeholder	Grey hint inside the empty field (up to 500 characters).
Required	Whether an answer is needed.
Text before / after field	Lead-in and trailing text on the same line. The Preview row below shows the result live.
Width (px)	40 to 1000, default 190.
Max chars	Up to 500, default 200.
Format	Checks the answer against a pattern (table below).
Alignment	Left, Center or Right.

Formats (input validation)

Multi-line options

Multi Line gives a text box with Height (Rows) (Small 3, Medium 5 as default, Large 8, Extra Large 10, Huge 15), Min Characters and Max Characters (up to 5000), Show character counter (off) and Allow user to resize (on).

11. Numeric Input

Numeric Input collects a single number and checks it against optional minimum and maximum bounds. You can wrap the field with lead-in and trailing text. The Pizza last_order_amount item reads "I spent [ ] €" with bounds 0 to 999, and age reads "I am [ ] years old".

Field	What it does	Default
Text before / after field	Lead-in and trailing labels ("I spent" ... "€").	Empty
Placeholder	Grey hint inside the field.	Empty
Min Value / Max Value	Values outside the range are rejected. Both within ±1,000,000,000; Min must not exceed Max.	No limit
Width (px)	40 to 1000.	150
Required	Whether a value is needed. Items created by dragging start with Required on.	Off
Allow decimals	Accept fractional values. Which decimal separator participants may type is a project setting (Project Settings, Display & Messages); stored values always use a dot.	Off

12. Constant Sum

A Constant Sum item asks participants to split a fixed total across several categories, which forces trade-offs. The Pizza importance_constant_sum item distributes 100 points across five aspects of ordering pizza.

Field / button	What it does
Import from TXT	Replaces the categories with the lines of a text file.
Min / Max per row	Optional lower and upper bound for that category (0 to 10,000).
/ Add Item	Remove a category or add one (up to 100).
Target Sum *	The required total, 1 to 10,000 (default 100).
Unit (optional)	Short unit next to the fields, for example points, % or EUR (up to 10 characters).
Checkboxes	Show running total (on), Allow decimal numbers (off), Required (on), Randomize items (off).

13. Custom HTML

The Custom HTML item inserts your own markup into the survey, for content the standard editors cannot produce, such as a specialized widget or an interactive layout. It is a power-user feature.

Editor fields and buttons

If the fields in your HTML and the declared variables do not match, Save asks first ("HTML fields and variables do not match") with Save anyway and Back to editor.

Storage: each value is stored under its declared variable name (up to 5000 characters). With no variables declared, the item is display-only like Text & Media.

14. Market Research

Market Research is a single item type that hosts three classic research methods. You add one "Market Research" item, then pick the method with a card under Research method. Each method has its own settings, its own participant experience, and its own analysis in the results. The Pizza study includes all three: KANO_Pizza, VanWest_Pizza, and GaborGranger_pizza.

14a. Kano, Prioritise features

Kano finds out how much each feature actually matters. For every feature the participant answers two questions: how they would feel if the feature were present (functional), and how they would feel if it were not (dysfunctional). The pair classifies the feature as Must-be, Performance, Attractive, Indifferent, and so on. The Pizza item tests five delivery features, one at a time.

14b. Van Westendorp, Acceptable price range

The Van Westendorp Price Sensitivity Meter asks four price questions about one concept: at what price it becomes too cheap, a bargain, expensive, and too expensive. Crossing the resulting cumulative curves yields a perceptual acceptable range and the historical OPP, where equal shares consider the price too cheap and too expensive. OPP is a minimum-resistance balance, not a revenue- or profit-maximizing price. The Pizza item prices a medium Margherita.

What the results show

• Price curves: the four cumulative curves and the acceptable price range (PMC to PME). This is the range in which respondents perceive the price as acceptable, not an optimal price.
• With a qualifying question: a tile Qualified, for example "240 of 400 (60%)". The price curves use the qualified respondents.
• With NMS: a second chart, "Purchase likelihood and revenue index (NMS)". Each answer is turned into a calibrated stated purchase likelihood with the weights 75% (definitely), 25% (probably), 10% (might), 0% and 0%. Between a respondent's prices the likelihood runs in straight lines: 0 at the too-cheap price, the bargain answer at the bargain price, the expensive answer at the expensive price, 0 at the too-expensive price. The revenue index is price × mean likelihood; the tile shows the price where this index is highest. With a qualifying question, the switch Base shows All respondents (those who did not qualify count with likelihood 0) or Qualified.
• Weights and variant: open the row Purchase likelihood weights on the results page to change the five weights (needs edit access); Reset to defaults restores 75/25/10/0/0. The switch Modified NMS keeps the bargain likelihood for all prices below the bargain price instead of falling to 0 at the too-cheap price (classic, the default). The weights are an assumption about how stated intent translates into buying; the result stays stated purchase likelihood, not a sales forecast.

14c. Gabor-Granger, Stated purchase intent & revenue index

Gabor-Granger shows one configured price at a time and asks whether the participant would buy at that price. In Choicery's sequential-adaptive mode, a "yes" moves to the next higher tested price and a "no" to the next lower one until the path reaches a ladder boundary or reverses at an adjacent level. The completed path records the highest accepted tested price, not an exact personal willingness-to-pay ceiling. Under the monotone reservation-price assumption, adjacent yes/no decisions bracket a switching interval; all-yes and all-no paths provide one-sided bounds. Across valid responses, Choicery estimates stated would-buy shares and a revenue index (price × stated share) at the configured levels. These are directional survey indicators, not sales or profit forecasts. The Pizza item tests six price levels from $7.90 to $17.90.

What the results show

• Demand curve: the share of respondents whose answer counts as buying at each tested price, and the revenue index (price × share).
• With the 5-point scale: a second line, "Weighted purchase likelihood", with the same weights as NMS (75/25/10/0/0, changeable under Purchase likelihood weights), and its own revenue index. A respondent who was not asked a price takes the answer given at the nearest asked price. Example with prices 19/29/39/49/59: a respondent answers "probably" at 39 and "might" at 49, so 19, 29 and 39 count 25% and 49 and 59 count 10%. The weighted line may rise between two prices; it is not smoothed.
• With a qualifying question: the switch Base shows demand among All respondents (default; those who did not qualify count as not buying) or among the Qualified. Example: 400 respondents, 240 qualified, 120 buy at 29: 50% of the qualified, 30% of all. The price with the highest revenue index is the same on both bases, the height is not.

Comparing groups in the results

On the Market Research results page (Analysis > Survey Analysis), Group by splits the selected item by a grouping variable: a nominal or ordinal question (as set in Variable Config), a variable added in the Data Editor, a Latent Class segment, or the experimental condition. Up to 12 groups are allowed. A comparison table then shows one column per group, a No answer column for respondents without a value, and All. Each cell shows the value, its 95% bootstrap interval and n. Show charts for switches the charts to one group; Gabor-Granger demand, NMS likelihood and Kano Better/Worse also get an overlay with one line or colour per group. The intervals describe the uncertainty within each group; this is not a significance test, and overlapping intervals do not prove that two groups are equal.

15. Evaluation

The Evaluation item presents a set of options as cards and asks participants to react to them: choose one, rate them, allocate points, or rank them. It has its own full-page editor rather than the standard item editor. The Pizza delivery_service_eval item shows three delivery offers and asks which one the participant would choose, with a "None of these" option.

The editor steps

The header shows the save state and the Save evaluation / Save changes button, which stays disabled until something changed. Live counters in the rail show how many options and characteristics you have.

16. CBC & MaxDiff (Conjoint Tasks)

The Conjoint section of the Add Item panel holds the choice-based tasks, the core of Choicery. They work differently from the survey questions above: instead of a single question, you design a set of attributes and levels, Choicery builds an experimental design, and each task shows the participant competing product profiles to choose between. Conjoint tasks always take their own pages, and their results drive the analysis tools (HB utilities, market simulation, willingness to pay, and so on).

Each conjoint type has its own attribute setup, task layout, and analysis, so each one has its own detailed manual. Open the manual for the type you are building:

Storage: conjoint responses expand into one export column per task, named from the task type and the task name you assigned. CBC writes conjoint_{task name}_t1, MaxDiff writes maxdiff_{task name}_t1_best and _t1_worst, DR-CBC and Best-Worst CBC add their _none columns, and fixed holdout positions are labelled FIXED_t{n}. The same responses are converted to long format for HB estimation. Each manual documents the exact schema and the analysis workflow.

17. Page Breaks

A Page Break starts a new page. In the editor it does not appear as a card but as the header of the next page (see Figure 1). Add one with New Page / Page Break in the Add Item panel (inserted immediately) or with Add Page at the end of the flow; remove one with Remove Page Break. Page breaks collect no data and are invisible to participants.

When to use them

• Between topic groups. Separate warm-up attitudes from demographics from conjoint tasks.
• After long questions. Keep pages manageable and avoid scroll fatigue.
• After instructions. Ensure participants read instructions before the questions appear.

18. End Pages & Outcomes

An End Page is a final screen, usually a thank-you message. Reaching it ends the survey. Every End Page sits alone on its own page. A survey can have several End Pages, for example one thank-you page for people who finished and one "Thank you, unfortunately you do not belong to our target group" page for people who were screened out. Skip logic can jump to a specific End Page. Items placed after the last End Page are never shown, and the editor warns about them.

Field	What it does
Thank You Message *	Rich text shown on the page (called Thank You Content in the add form). Prefilled with "Thank you for your participation! Your responses have been successfully recorded. You may now close this window."
Outcome	How a response that reaches this page is counted: Complete (default), Screened out or Quality fail. Use the last two for pages that end the survey early.

What the outcomes mean

Screened out, Quality fail and Quota full responses never enter the analysis, the dropout figures or the quota counts. The Analysis overview shows them as separate tiles ("Screened", "Quality", "Quota full") and lists them below the funnel as "Ended by the survey and not counted as dropoffs". On an End Page with outcome Screened out or Quality fail, participants see the heading "Survey Ended" instead of "Survey Completed". The preview of an End Page says how respondents reaching it will be marked.

Storage: the outcome is saved with the response and exported in the column outcome (or response_outcome if a question is already called outcome), with the values complete, screened, quality or quota_full: <quota name>. The column sits in the system block at the right of the export; it is empty for older responses recorded before outcomes existed.

19. Quotas & Panel Redirects

A quota limits how many completed responses each answer of a single-choice question may collect, for example 150 men and 150 women. Once a group is full, further participants in that group end the survey. Quotas are not set in the survey editor but in the project settings: top navigation Settings, section Fieldwork, entry Quotas (between Availability and Privacy).

Setting up a quota

How quotas are counted

• Places taken in a group = completed (not deleted) responses with that answer, plus participants who are currently in the survey with that answer. A place held by an unfinished participant expires after 2 hours. Screened out, Quality fail and Quota full responses never take a place.
• A quota added later also counts the completed responses collected before it existed.
• When it is checked: when the participant clicks Next on the page with the quota question (before skip logic runs), and again when the survey is completed. If the question is hidden by display logic, no limit applies to that participant.
• Participants already in the survey keep the quotas from the time they started. A deleted quota stops applying at once; for a changed target, the more generous value applies to them.
• Quotas are not checked in Preview, so test them with the live link.

What a participant in a full group sees

The survey ends with the outcome Quota full. The participant sees the Limit Reached Message from Settings > Availability (default: "We have already received the required number of responses. Thank you for your interest!"), or is sent to the Quota full panel link if one is set. The overall limit Maximum Responses (Settings > Availability > Response Limit) works the same way and is recorded as quota_full: Maximum Responses.

Panel redirects

When your participants come from a panel provider, you can send them back to the panel at the end. On the Publish page, the card Panel Redirects has one link per outcome: Complete, Screened out, Quota full and Quality fail. Write {panel_id} where the panel expects the participant's ID; Choicery fills in the ID it received in the survey link. A field left empty shows the End Page text (or the Limit Reached Message) instead. Save with Save Panel Redirects; like quotas, the links apply at once without publishing again.

Links must start with https:// or http://, may be up to 2,000 characters long and allow {panel_id} as the only placeholder, after the domain name. When a panel link is set for an End Page's outcome, the End Page editor notes that participants are sent there; in Preview and test mode the End Page shows the link with TEST as the ID instead of redirecting. Someone who reopens the survey after finishing sees the local page, never a second redirect.

20. Display & Skip Logic

Two related features branch the survey based on earlier answers. Display Logic shows or hides a single item; Skip Logic jumps participants forward over whole pages. Both are configured on the item card itself, through the Add Display Logic and Add Skip Logic links in the card footer (Figure 1, numbers 10 and 11). The Skip & Display Logic entry in the Survey Management panel is a read-only overview of the rules you already have.

How Display Logic works

How Skip Logic works

Skip Logic sits on the question that triggers the jump. Click Add Skip Logic in its card footer, switch on Enable Skip Logic, choose ALL (AND) or ANY (OR) in the "Skip when ... conditions met" row, and give each condition a target. After answering, matching respondents jump ahead to the page containing the target; every page in between is skipped. The target can be any item on a later page, a specific End Page, or End Survey (Skip to end), which always ends as Complete (section 18).

Condition types

21. Best Practices

Survey design

• Keep it short. Every extra question lowers completion. Cut anything not essential to your analysis.
• Use clear language. Avoid jargon, double negatives, and asking two things at once.
• Put easy questions first. Warm up with simple attitudes before demanding tasks like conjoint or Market Research pricing.
• Group related questions and separate topics with page breaks.
• Randomize where it helps. Shuffle options on choice questions and rows on grids to reduce order effects.

Technical tips

• Name variables meaningfully. Use names like diet, food_involvement, or last_order_amount, not the auto-generated IDs. These become your export headers.
• Set validation. Use min/max on numeric fields, Text Input formats for emails and dates, and selection limits on checkboxes. Validation catches bad data at the source.
• Give every ending its outcome. Screen-out and quality-check exits belong on End Pages with the matching outcome, so they never mix with your completes.
• Test before launching. Preview the whole flow, walk every logic branch, test quotas with the live link, and check that a few test responses export correctly.

Data quality

• Add an attention check. A question with a known answer ("Please select Somewhat Agree") flags inattentive respondents; send failures to an End Page with the outcome Quality fail.
• Watch completion times. Flag responses far below the median. The duration_seconds field makes this easy.
• Soft-launch first. Collect 10 to 20 responses and review the data before opening to the full sample.

22. Quick Reference Table

Every item type, its internal key, storage pattern, and data type.