---
name: survey
description: Plans a short survey from what someone wants to learn and from whom, writes its questions in the order that gets them finished, and builds the survey page in one of Fil's survey styles on Fil's survey kit, so the answers are kept in Fil once it's published. Use for customer, client, waitlist, product, event and research surveys, feedback forms and questionnaires.
---
# Survey

## Read first
Brief.md: who's asking, the goal, who answers and how they'll reach it, the decisions the answers drive, anything that must be asked, the picture question, the price question, the voice, the style, the email and the thanks. Questions.md: the survey's words and every question. The pictures wired in are the picture question's choices, each named after its choice; a logo wired in goes at the top. A picture of a survey or a form (the style picked, or one whose look they like) only sets the style: use Brief.md's style, or the kit's style closest to that picture, and never show it.

## The questions
- Ten to twelve at most, about three minutes to answer. Each earns its place by a decision Brief.md names: cut any whose answer wouldn't change what they do.
- No more than three are required: the ones the decisions can't be made without, never how they heard or whether they may be followed up. Everything else can be skipped.
- Each is one of four kinds: pick many, pick one, a few short lines, or a paragraph. A pick may end with "Something else", which opens a line for their own words.
- The order:
  1. Quick, inviting taps first: what interests them (with pictures when the choices are things people can see), who they are, what they use now.
  2. The one required typed question once they're invested (a few lines: name up to three projects, places, ideas), never first.
  3. How often, the price, what they'd pay more for and support, together and in that order.
  4. The open "what would make it a must-have" (a paragraph) right after the money questions.
  5. Quick taps last, right before Send: how they heard, and whether they may be followed up.
- The price question's four choices span a real range, each step about double the last (e.g. $20, $50, $100, $200 a month), with "Less than" the first and "More than" the last beside them.
- Tools and options are grouped by job, never one brand each: "Figma, Canva or Photoshop", "After Effects, Premiere or CapCut".
- Questions in plain words to "you", at most twelve; choices of two to five words, with a short note only where one helps; a hint only when it helps them answer. Choices are ones this audience would recognise, three to eight of them, and never overlap in a pick of one.

## The page
One HTML file on Fil's survey kit, which draws the survey in its style, ticks and checks the answers, keeps them in the browser until they're sent, and sends them to Fil once the page is published, where its owner reads them by question:
```html
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1">
<title>The title · the brand</title><meta name="description" content="The intro">
<link rel="stylesheet" href="https://fil.design/kits/survey/1.css">
</head>
<body>
<main data-fil-survey="paper"><h1>The title</h1><p>The intro</p></main>
<script type="application/json" data-fil-form="survey">{ "title": "…", "questions": [ … ] }</script>
<script src="https://fil.design/kits/survey/1.js"></script>
</body>
</html>
```
- data-fil-survey is the style Brief.md names, as the kit calls it: paper (Paper), swiss (Swiss), cafe (Café), soft (Soft), brutal (Brutal), gilded (Gilded), ticket (Event pass), terminal (Terminal), studio (Studio), chat (Chat).
- The JSON, from Questions.md: { "title", "label" (the small line), "brand" (the name, in words), "logo" (the wired logo's address, when there is one), "intro", "start": "Begin", "sendTitle" (the line above Send), "send": "Send", "email": { "place": "you@company.com (optional)", "required": false } only when Brief.md asks for one, "thanks": { "title", "text" }, "questions": [ … ] }.
- A question: { "id" (short, lowercase, its own), "kind": "many" | "one" | "texts" | "text", "look": "pictures" | "rows" | "price" | "pills", "required": true (only the required ones), "label" (one or two words), "title" (the question), "hint", "options": [{ "id", "label", "note", "picture" }], "other": "Something else", "places": [three examples] for a few lines, "place" for a paragraph, "unit": "a month" for a price }.
- Looks: pictures for choices people can see, price for the price question, rows for choices with notes or long labels, pills for short ones. The price question's four steps are its options, each labelled with the amount alone ("$500") and the period in its "unit" ("a month"), its "Less than…" and "More than…" options with "aside": true.
- A picture choice's "picture" is the wired picture named after that choice, its address written exactly as attached (facet:asset/…), matched by name and what it shows, never by the order they come in. Without its pictures, that question's look is rows.
- The words exactly as Questions.md writes them, and the JSON valid. Nothing else in the page: no other script, style or markup; the kit draws all of it.

## The picture question's pictures (4:3)
Each choice's picture shows that choice alone, plainly, as the person answering would picture it: one subject, centred, in soft even light, every picture of the set in the same look and light so none stands out. No words, logos or interface in them.

## Avoid
More than twelve questions or three required, a typed question first, leading or two-in-one questions, jargon, a price range too narrow to learn from, tools listed one brand at a time, and choices the audience wouldn't recognise.