Loop & Merge repeats one question once per option somebody picked — or once per grid row they rated a particular way — and pipes that item's own wording into the question. Instead of one vague comment box at the end of a section, every verbatim arrives already attached to the thing it is about.
Say you are running an employee engagement survey. One grid asks people to rate seven statements about recognition and pay from Strongly disagree to Strongly agree. Underneath it, you want to know why— but a single “anything you'd like to add?” box will not tell you which of the seven statements a comment refers to.
One open text box: “Anything you'd like to add?”
You get one comment, if you get one at all. “This has been a problem for a while” could be about pay, about your manager, or about something else entirely.
One open text box per statement they rated down: “You said you disagree with: {{LOOP}}— would you like to tell us a little more about what's behind that?”
Someone who rated down two statements answers twice, each time about a named statement. Someone who rated nothing down never sees the question.
Loop & Merge lives in the Pipe panel, not the Logic panel. Logic is for showing and hiding whole questions; Pipe is for pulling earlier answers into this one.
A loop can only read a question that comes before it. Add or drag your follow-up question directly beneath the grid or checkbox question it should loop over.
Click the small Pipe link at the top-right of the question's title field. Scroll to the amber “Loop over selections” section at the bottom of the popover.
Under “Repeat for each selection from”, pick the question whose answers should drive the repetition.
When the source is a grid, a second dropdown appears: “Only loop rows where column equals”. Leave it on “any column” to ask about every rated row, or pick a single column to ask only about those.
Click Apply loop. The picker is replaced by an amber confirmation box, and a “Loops over Q…” chip appears in the question card header.
Click “Insert {{LOOP}} in title”, then write your question around it. The token is replaced at display time with the option or row the respondent is answering about.
A required loop question is required for every iteration. Somebody who rated six statements down would have to write six answers before they could submit. On sensitive subjects that is the fastest way to lose the response altogether.
Any of the question types below, as long as it appears earlier in the survey.
| Question type | Loops | Notes |
|---|---|---|
| Checkboxes | Once per ticked option | The most common source. Ideal for “which of these did you use?” followed by a question about each one. |
| Matrix / Grid | Once per rated row | Can be narrowed to rows rated with one particular column — see filtering below. |
| Image Choice | Once per selected image | Useful for concept and creative tests where each shortlisted design needs its own diagnostic. |
| Ranking | Once per ranked item | Iterates in the order the respondent ranked them. |
| Multiple Choiceonce | Once, for the single answer | Allowed, but a single-pick source loops exactly once. Text piping is usually the simpler tool. |
| Dropdownonce | Once, for the single answer | Same as Multiple Choice — the editor warns you when you pick a single-select source. |
This is what makes loops useful on agree/disagree batteries. Without a filter, the follow-up repeats for every row the respondent rated — including the ones they agreed with. With a filter, it repeats only for rows rated with the column you name.
The filter is an exact match against a single column label. There is no “contains”, no “either/or”, and no free-text entry — you pick one of the grid's own columns from a dropdown.
So on a five-point scale, a loop filtered to Strongly disagree will not pick up people who chose Disagree.
Add a second copy of the follow-up directly beneath the first, with identical wording, and set its filter to the other column. Respondents cannot tell the difference — they simply get one follow-up per row they rated down. The survey only gets longer for the people who actually rated something down.
If you find yourself wanting three or four columns covered, that is usually a sign the scale itself should be narrower — or that the follow-up belongs on fewer statements.
Works in both the question title and the description, and can appear more than once in either. It is replaced with the option label or grid row wording for the iteration being answered — which is why grid rows should read as complete statements. A row phrased as a question will read oddly once it is piped into a sentence.
If the source is unanswered, or the filter matches none of the respondent's rows, the question is skipped entirely rather than shown empty. Because hidden questions are not validated, this happens cleanly even on a required question — though we still recommend leaving Required off.
Each iteration is stored as its own answer, labelled with the option or row it belongs to. In a CSV export you get one column per possible item, filled in only where that respondent answered — so you can sort every verbatim by the statement that prompted it without any manual coding.
Carry-forward takes one question and fills its options, rows or columns from an earlier answer — one question, one answer. A loop renders the same question several times over — one question, several answers. Reach for carry-forward when you want to narrow a list, and for a loop when you want to ask the same thing about several items in turn.
Why: The source has no selections, or the column filter matches none of the rows the respondent rated.
Fix: Answer the source question first, and check the filter matches a column label exactly as it is spelled in the grid.
Why: The token is in the title but the question is not actually looping.
Fix: Re-open the Pipe panel and confirm the amber “Looping over Q…” box is present. If you see the picker instead, the loop was never applied.
Why: The loop is configured but the title has no {{LOOP}} token, so there is nothing to substitute.
Fix: Click “Insert {{LOOP}} in title” in the amber box. The editor also shows a reminder on the question card when this happens.
Why: Loop sources must come before the looping question, and must be a supported type.
Fix: Drag the follow-up question below its source, or change the source to a supported type.
Why: You picked a Multiple Choice or Dropdown source.
Fix: That is fine if intended. If you expected several iterations, the source needs to be a Checkbox, Image Choice, Ranking or Matrix question.
Once answering the source so the loop fires — check the iterations appear and the wording substitutes properly. Then once without triggering it, to confirm the question disappears cleanly and the survey can still be submitted. A loop that silently fails to appear looks identical to a loop that was never set up.