Studio Projects

Projects

A project is one study: its questionnaire, its analysis flows, its settings, its response database, its deployments and its complete version history. This page covers what a project is made of, how Saves and drafts work at a glance, the Projects list, creating a project (blank, from the example study, from a template or from your organization's library) and what Studio sets up for you when it does.

---

What a project is made of

A project is a handful of documents plus a database. You edit the documents through the screens; they matter because they are what gets versioned, downloaded and reproduced.

Document	What it holds	Edited in
survey/questionnaire.json	the questionnaire: pages, blocks, questions, logic, quotas, codebook, theme, scripts	Builder
flows/<name>.flow.json	one analysis flow each: nodes, connections, parameters, live tiles	Flows
studio/settings.json	environments, runtime, connectors, pinned insights, report house style, study and citation metadata	Settings, Data, Flows
analysis/<name>.codeframe.json	a coding scheme (codeframe) for one open-text question: its themes, their word rules and the answers coded by hand	the codeframe editor, from Flows or Files — see Coding Open Answers

The Builder's Source tab shows the questionnaire document as JSON and lets you edit it directly — see The Source tab.

Alongside the documents the project owns:

• a response database — the responses table (one row per interview), quota_counters (how full each quota cell is), survey_meta (the title and variable dictionary of each published survey) and any table your flows write;
• files — media and data you upload, and the outputs of every flow run (Files);
• deployments — each published version of the survey, per environment (Publishing and Environments).

---

Saves and drafts

Save creates a numbered version of the whole project (#17): every changed document gets a new version, the engine validates it and generates its Python (questionnaire.py, <flow>.py), and the set is stamped with a number, an author, a message and a validation state. A Save message is optional (up to 240 characters); when you leave it empty Studio writes one, such as Update questionnaire or Update questionnaire, tables.

The Save badge next to the project name in the topbar shows the current state from every tab. Click it to open that Save in History.

Badge	Meaning
● valid #17	Save 17 is clean — publishable and runnable
● warnings #17	saved; the engine has remarks — publishable after you confirm. The remarks can include findings the engine grades as errors (an empty page, a design that cannot be estimated); they do not block publishing, but the confirmation names them
● errors #17	saved but broken — cannot be published or previewed
● checking #17	the Save's validation has not reported yet
saving…	a Save is in progress
unsaved	the project has never been saved

The dot is green, amber or red for valid, warnings and errors.

Drafts. Between Saves your edits are autosaved as a private draft, 1.5 seconds after your last change, so a closed tab costs you nothing. A draft is not validated, not published and not visible to colleagues — except as the live document a colleague follows while you hold the edit lock. See Saving and Drafts and autosave.

Publishing pins a Save, flows run against a Save, and any Save can be restored or downloaded as a research bundle. The full story is in History and Versions.

---

The Projects list

The organization's Projects tab lists every study in the workspace.

┌ Projects   Acme Research                                        [+ New project] ┐
│ [Search projects by name…           ]  [Sort: Last opened ▾]                     │
│ ┌──────────────────────────────────────────────────────────────────────────────┐ │
│ │ Name                      Version  Status      Responses · 14d  Updated      │ │
│ │ Employee Pulse 2026 Q1    #17      ● valid     ▁▂▅▇▆  247       Jun 4, 2026  │ │
│ │ Brand tracker (pilot)     #3       ● warnings  ▁▁▂▁▃   31       May 28, 2026 │ │
│ └──────────────────────────────────────────────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────────────────────────┘

Column	What it shows
Name	the project's name; click the row to open the project on its Builder tab
Version	the number of the latest Save (#17), or unsaved
Status	a validation label: valid, warnings, error or unsaved
Responses · 14d	total responses collected, with a sparkline of the last 14 days (shown for the first 20 projects)
Updated	a date for the project

Note. Status and Updated are a quick guide, not a live record: they are not always current for projects you have not saved in this browser session. For the authoritative state, open the project and read the Save badge or History.

• Search — the box (placeholder Search projects by name…) matches the project's name or its slug. With no match you see No matching projects and a Clear the search button.
• Sort — Sort: Last opened (the projects you opened most recently in this browser first), Sort: Name (A–Z) or Sort: Date created (newest first).
• Empty organization — No projects yet — "Each project is a questionnaire, its flows and its own response database. Create it, preview it as a respondent, then Save and deploy." — with two buttons: New project and Start from the example study. For a member (not an owner or admin) both are disabled, with the line "Only owners and admins can create projects — ask one to set it up."

The workspace chip in the topbar (organization / project name) also lists up to ten projects, plus All projects and New project. Switching project from there keeps you on the same tab. New project in the chip is disabled in the same cases as the button on the list: at your plan's project cap, and for members.

Project caps

Plan	Projects per organization
Free	2
Plus	10
Pro, Corporate	unlimited

At the cap every New project button — on the list, in the workspace chip, Start from the example study on an empty list, and the New project buttons on the Library tab (on each template card and on each saved questionnaire) — is disabled; hovering says "Your plan allows 2 projects — upgrade to add more" (10 on Plus). The list also shows, for example on Free: "You've reached the 2-project limit on the Free plan. Upgrade your plan to add more." See Plans, Trial and Billing.

Plan. On the Free plan each project also takes at most 1,000 completed responses, counted over all of its environments together. Screen-outs and partial responses do not count toward it, and neither do the sample responses of the example study.

---

Creating a project

Plan and role. Only owners and admins can create projects. For a member, every New project button — on the Projects list, in the workspace chip and on the Library tab — is disabled ("Only owners and admins can create projects" when you hover it), and the Library tab adds under its templates: "Only owners and admins can create projects — ask one to set it up." Members can open, edit and Save every project in the organization.

1. On Projects, click New project (or Start from the example study on an empty list, or New project on a template or saved questionnaire in the organization's Library).
2. Type a Name. Under it Studio shows the address the project will have: "Its address will be /brand-awareness-study-2026".
3. Under Start from, choose Blank project or Template (see below).
4. Click Create. The button reads Creating… while Studio sets the project up, then the project opens on its Builder tab. The toast says Project brand-awareness-study-2026 created (for the example: Example study created — open the Builder to explore it).

Enter in the Name field also creates. Create stays disabled until there is a name and, if you picked Template, a template.

The New project dialog

┌ New project ──────────────────────────────────────────── ✕ ┐
│ Name                                                       │
│ [Brand Awareness Study 2026                             ]  │
│ Its address will be /brand-awareness-study-2026            │
│                                                            │
│ Start from                                                 │
│ ┌────────────────────────────────────────────────────────┐ │
│ │ ○ Blank project                                        │ │
│ │   Start with one placeholder question                  │ │
│ ├────────────────────────────────────────────────────────┤ │
│ │ ● Template                                             │ │
│ │   [Course evaluation                               ▾]  │ │
│ │   End-of-course feedback: the course, the teaching, …  │ │
│ └────────────────────────────────────────────────────────┘ │
│                                        [Cancel] [Create →] │
└────────────────────────────────────────────────────────────┘

• Blank project — "Start with one placeholder question".
• Template — "Start with a pre-built questionnaire". Selecting it shows a picker (Choose a template until you pick one). The picker opens a list with a search box (Search 13 templates… — the count includes the example study and anything your organization saved) and, per row, the name, its size (8 pages · 12 questions) and a one-line description. If nothing matches: "Nothing matches “…”." When your organization has saved questionnaires, the list has two headings: Templates and Your organization's library.
• Picking a template while the Name field is empty (or still holds the name of the previous pick) fills in the template's name; once you type a name of your own, changing the template leaves it alone.
• Keyboard: ↓ opens the list, ↑/↓ move through it, Esc closes the list (a second Esc closes the dialog).

The project's slug

The slug is the project's permanent address in Studio (studio.siamang.org/<org>/projects/<slug>). It is derived from the name and cannot be changed later — not even by renaming the project.

How the name becomes a slug:

1. The name is spelled in Latin letters: Cyrillic (Russian, Ukrainian, Belarusian, Kazakh, Serbian, Macedonian) and Greek letters are transliterated, and accented Latin letters lose their accents (é → e, ß → ss). Characters with no Latin spelling (Chinese, Japanese, Arabic, Hebrew, …) are dropped.
2. It is lower-cased, every run of characters other than a–z and 0–9 becomes a single hyphen, and hyphens are trimmed from the ends.
3. A slug longer than 64 characters is cut to 64, at a word break when there is one near the end.
4. A name with nothing left to spell (only CJK characters, or only punctuation) becomes project; one or two characters get -project added.
5. If the organization already has a project with that slug, -2, -3, … is added.

Name	Address
Employee Pulse 2026 Q1	/employee-pulse-2026-q1
Опрос удовлетворённости	/opros-udovletvorennosti
Щедрий вечір	/shchedriy-vechir
Café Größe	/cafe-grosse
Q1	/q1-project
調査	/project
a second Опрос	/opros-2

The dialog shows the result under the name ("Its address will be …") before you create anything, so what you see there is the address the project gets. The slug is always 3 to 64 characters of lower-case letters, digits and hyphens, starting and ending with a letter or digit. You can rename the project to anything afterward; the slug stays.

Blank project

One page (page1) holding one placeholder question, and default settings. The question is an optional Single choice called "New question", with the options 1 Option 1 and 2 Option 2, Id and variable q1, and its codebook entry. Rewrite it into your first question or delete it, then add the rest — see The Builder.

The placeholder is there so that the new project checks clean: a page with nothing on it is an error in the engine's check (EMPTY_PAGE). If you delete the question and leave the page empty, that finding comes back until you add a question or a Body.

The example study

Example study is the first entry in the template list ("the full Digital Life & Wellbeing study — questionnaire, six analysis flows and sample responses"). It is the fastest way to see what Studio does with a real study: a questionnaire that uses every question type but one, 729 sample responses already in the database, and six analysis flows that clean, tabulate, test, model and segment them, each with a report of its own. Whether you start it with Start from the example study on an empty list or with Template → Example study in the dialog, the name offered is Digital Life & Wellbeing (example); the questionnaire keeps its own title, Digital Life & Wellbeing 2026. Everything in it — questions, labels, flows and reports — is in English.

The study asks how everyday technology use relates to how people feel, then tests an idea for an app that helps people get the balance right between their phone and the rest of their day.

The example's questionnaire

22 pages (a respondent sees at most 17 of them before the page the interview ends on), 33 questions and 65 variables, about ten minutes to answer:

Page	What it asks or shows
About this study	an introduction, no questions
Your consent	"Do you agree to take part?" — Yes, I agree / No, not now as buttons, required
No problem	a Screen-out page with Show if consent = 2, so only those who decline see it
A quick check	age in years (required, valid range 13–99)
Thanks for your interest	a Screen-out page with Show if age < 16
About you	gender (with 99 Prefer not to say), age group (required — asked on purpose beside the age, so that cleaning can check one against the other) and type of area (a dropdown)
Work & study	the current situation (required)
Where you work or study	Show if employment ≠ 5 (everyone but those not working): how they mostly work or study, and the one-way commute in minutes — asked only of those not fully remote, a question-level Show if with AND and NOT
Your devices	two blocks in random order: the devices they own (a wide Multiple choice, one yes/no variable per device) and the one they use most, whose Smartwatch option is shown only to smartwatch owners (an option-level Show if)
Apps & media	two blocks in random order. Social, whose two questions are also shuffled: how often they use social media, and "On social media, I scroll for longer than I intended" with Not applicable. News: their main source of news, its options shuffled by a Shuffle options script, and whether they trust the news they see, asked only if that source is social media, news apps or TV / radio (news_source in 1, 2, 3)
The apps you use	the kind of app they spend the most time on (options shuffled), then how a long session on it leaves them feeling — eight feelings, one yes/no variable each — with the kind of app piped into the question as {label:main_app}
Screen time	hours a day on screens (typed, in whole hours), and what they do to manage their screen time, with an exclusive None of these
How you feel	a matrix of five statements rated Never … Always ("I wake up feeling rested", "I sleep well", …), life satisfaction on a 1–7 scale, and a ranking of what helps them disconnect
You and your phone	a matrix of seven agree–disagree statements about the phone; the fourth is an attention check ("To show you are reading, please choose “Disagree” here")
An app idea (two pages)	an A/B message: one page frames the idea around sleep, the other around time. An Assign to a condition script draws each respondent's arm into message_arm before the first page, and each page has Show if on it
The app	interest in trying the app (0–10) and a Best–worst (MaxDiff) question on eight features: six tasks of four, in 20 versions
Using the app	the features they would actually use (None of these exclusive), and how likely they would be to recommend the app (0–10, a Net Promoter question)
What it should cost	four Van Westendorp price questions in $ a month, on a page with Show if app_interest ≥ 3
Almost done	how likely they are to try a digital detox (0–10), one change they would make to their digital life (open text, optional), and whether they would like to see how the survey was built
Taking you there	a Redirect page, for those who said yes, to the open-source toolkit's page (https://github.com/Siamang-Labs/siamang)
Thank you	the Final page

Along the way it uses:

• Every question type except Conjoint: Single choice as radio buttons, a dropdown and buttons; Multiple choice in both layouts (a yes/no variable per option, or one variable holding the list, with an exclusive "None of these"); Likert scale with 5, 7 and 11 points; Number (age, commute, hours, prices); Open text; two Matrix questions; a Ranking; and a Best–worst (MaxDiff). Every question lists its own options, items or rows, so the Builder shows all of them and Validation finds nothing to fix. Most are written from the codebook's value labels, so the two cannot drift apart; the wide multiple choices and the matrices word theirs for the respondent ("A smartphone", "I sleep well") and keep short labels for the tables ("Owns a smartphone", "Sleep well").
• Logic at page, question and option level, with AND, NOT and in, piped text, and an A/B message.
• Randomization of blocks on two pages, of the questions in one block and of the options of four questions, plus the Shuffle options script.
• Three quota cells on age group, 400 each.
• A codebook with declared missing values: gender 99 = Prefer not to say (a refusal) and, on the scrolling statement, -1 = Not applicable — the code the survey stores when a respondent picks Not applicable. Life satisfaction's labels are what the respondent sees: 1 Not at all, 2 … 6, 7 Completely.

The example's sample responses

729 synthetic responses are in the database when the project opens, so Data, Live, Flows and Reports have something to show before any fieldwork: 720 interviews, dated across the ten whole weeks before the week you created the project (about a quarter of them in the first week, then a steady trickle), and 9 of them submitted a second time. 674 are completed — 258 of those left through the redirect page — and 55 are partial interviews that broke off on the How you feel page or later. They are tagged with the survey id sample-data, and they never count toward a response cap: on the Free plan the project still takes 1,000 completed responses from real respondents. Every example project gets the same answers — only the dates move with the day you create it — so its reports show the same numbers.

The answers are drawn from a model, so that every analysis in the flows has something to find:

• Who answered. The sample is skewed the way online samples often are — too young, too many women, too many city centers — which the cleaning flow's weights correct.
• Three kinds of user, Balanced, Always on and Intentional, who differ in hours on screens, social media use, feeling in control and restlessness without the phone, and whose shares differ by age group.
• Two scales. The five How you feel statements share one wellbeing factor (anxiety runs the other way). The six phone statements hold two ideas: the pressure the phone puts on people (replying at once, comparing, restlessness) and what it gives them (closeness, community, an easier life).
• What goes with life satisfaction. Sleep first, then feeling in control and calm; what the phone gives helps and its pressure hurts; age adds a little.
• A trend. Hours on screens drift down by about 0.12 hours a week over the ten weeks.
• Apps and feelings. Each kind of app leaves its own mix: short videos entertain and leave people tired and feeling they wasted time, messaging apps connect, news apps inform and unsettle, streaming relaxes.
• The app idea. The eight features have a clear order of importance, with differences by age group and by kind of user; the features people would use gather around three needs (rest, focus, company), so a shortlist that spans them reaches the most people; the four prices center on about $3.20 a month; and the message about time raises interest more than the one about sleep, least among the 45+.
• Problems to clean. Break-offs, speeders, straightliners, attention-check slips, age groups that contradict the age typed, the answers submitted twice, and a few commutes typed in hours as minutes (300 to 600, outside the codebook's valid range of 0–240).

The open answers to "If you could change one thing about your digital life, what would it be?" are coded by a codeframe the project ships, analysis/improve.codeframe.json — six themes, and a theme and a tone for each answer it knows (see Coding Open Answers). Two answers it has never seen are left uncoded, so its coverage stays short of 100 %. It is an older, version 1 codeframe, which the flow tables applies (node code, Also add sentiment on). Edit codeframe… on that node, or Files → Codeframes → edit, opens it in the codeframe editor, where you can code those two answers by hand or give the themes word rules; its first Save there makes it version 2.

The example's flows

Six flows, numbered in their titles in the order Run all runs them, each with a report of its own:

Flow	Title	What it does	Report
cleaning	1. Clean raw responses	checks the completed interviews against the codebook; drops repeat submissions, speeders (under four minutes) and failed quality checks; weights to illustrative population shares of age group, gender and area; writes the table clean_responses and an R bundle	Data quality
tables	2. Key tables	weighted frequencies, a crosstab and a chart of life satisfaction by age group; a banner table of habits by age group and gender with significance letters; the open answers coded; a tab book in Excel	Key tables, with its tables in Excel
usage	3. Screen use	hours on screens by age group, the weekly trend over the fieldwork, the ways of managing screen time compared (Cochran's Q), and a perceptual map of the kinds of app and the feelings they leave	Screen use
wellbeing	4. Wellbeing: scales and drivers	a wellbeing index and its reliability, two factors in the phone statements, the key drivers of life satisfaction and an ordinal regression; writes the table scored_responses	Wellbeing
wellbeing_app	5. The app: features, reach and price	MaxDiff of the features, TURF of the ones people would use, the Net Promoter Score, Van Westendorp prices, and the A/B message's effect on interest (Welch's t-test)	App features and price
segments	6. Segments	three segments by k-means, who they are, how they feel and what each wants from the app, and the MaxDiff choices saved for a hierarchical Bayes estimate in R	Segments

The other five flows read the table cleaning writes — segments through the one wellbeing writes from it — so run cleaning first, or all six at once with Flows → More ▾ → Run all flows; the reports then appear on Reports, with the combined report titled Digital Life & Wellbeing 2026. What each flow shows you, node by node, is in The example study's flows.

Plan. The six flows come with the project on every plan. The Free plan allows 3 flows per project: there you can run, edit, rename and delete the example's flows, but a Save that adds a flow is refused until fewer than three are left.

What else the example sets up

• Five pinned widgets on Data → Insights: overall life satisfaction, life satisfaction by age group, hours a day on screens, would recommend the app, and interest in the app by message. Like all of Insights they count every row, partial interviews included, unweighted — the flows' reports read the cleaned, weighted data and give other numbers.
• A report house style (Settings → Reports), the Look of all six reports: Typeface modern, Tables zebra, teal Links and Chart colors of teal, orange, indigo, sky, olive and plum, which readers with protanopia or deuteranopia can tell apart too. Every chart of the six flows has Palette theme, so it is drawn in those colors. The combined report uses the house style, and a new Save report node starts with it.
• Live tiles. The cleaning flow has Live: recompute on new responses on, with the tiles Clean respondents and Quality checks; the usage flow has a tile Screen time by week — its report's weekly trend — with Live off. They appear on Live once the flows have run. See Live Monitoring.

Built-in templates

Twelve complete, publishable questionnaires. All of them validate cleanly as they are. Most start with a consent page (Yes/No buttons, with a screen-out page for people who answer No) and end with a thank-you page; rename the brands, the course or the product and go.

Template	Description	Pages · questions · variables
Brand awareness tracker	Awareness (aided and unaided), consideration, usage, brand image, NPS and demographics — the classic tracker, ready to rename the brands.	8 · 12 · 18
Employee engagement pulse	Engagement, manager, workload and eNPS in five minutes, with the group cuts kept to ten people or more.	8 · 8 · 15
Course evaluation	End-of-course feedback: the course, the teaching, an overall rating and two open questions.	6 · 8 · 15
Customer satisfaction (CSAT)	Transactional feedback right after a support contact: the channel, satisfaction, effort, recommendation and the demographics to cut by.	7 · 11 · 11
Product feedback	How often the product is used, satisfaction with each part of it, the one missing capability and a price that would feel fair.	7 · 7 · 11
Website experience (UX)	What the visitor came to do, whether they managed it, a four-item usability scale and the blocker in their own words.	5 · 6 · 9
Event feedback	Overall rating, the program, the practical side and whether they would come back next year.	7 · 6 · 11
Concept test	One concept on its own page, then purchase intent, how the idea is judged, two price points and demographics.	6 · 8 · 11
Module evaluation	The module as a unit of study: intended outcomes, prior knowledge, real workload in hours and how it was assessed.	7 · 7 · 12
Health and wellbeing check	Self-rated health with a refusal code, a five-item wellbeing scale, activity and sleep — with the consent wording a health survey needs.	7 · 9 · 13
Pilot study	The shape an ethics committee expects: information sheet, two-part consent, an eligibility screener, a short instrument and a debrief page.	7 · 4 · 13
Exit interview	Why someone is leaving, what would have kept them, and whether they would come back or recommend the place.	7 · 8 · 12

The new questionnaire takes the project's name as its title. The same list, with a New project button on each card, is on the organization's Library tab.

From your organization's library

Questionnaires your team saved with Library → Save questionnaire as template… in the Builder appear in the same picker under Your organization's library (with their description, or "a questionnaire from your organization's library"). The new project gets a copy, retitled with the project's name; later changes to the library item do not reach it.

Plan. Saving to the organization's library is (Plus). Starting a project from a questionnaire that is already in the library works on every plan. See Question Bank, Templates and Library.

---

What gets created

Creating a project sets up, in one step:

1. A response database for the project.

2. Save #1, with a message that says where it came from:

   Started from	Save #1 message
   Blank project	Initialize empty project
   Example study	Initialize example study (Digital Life & Wellbeing)
   A built-in template	Initialize from template “Course evaluation”
   A library questionnaire	Initialize from library questionnaire “<name>”

3. Your organization's survey house style — logo, colors, typeface and the other look settings an owner set for the organization — copied into the new questionnaire's theme. It fills only what the starting questionnaire does not set itself, and it is a copy: changing the house style later does not touch existing projects. See Theme and Branding.

4. Two environments: pilot with a cap of 50 responses and main with a cap of 1,200 — counted in completed interviews, so screen-outs and partial responses do not use them up. See Publishing and Environments.

5. For the example study only: the six flows, the codeframe of the open answers, the 729 sample responses, the pinned insights and the report house style described above.

A blank project, like the example study and the built-in templates, starts at ● valid #1: its placeholder question keeps the engine's check clean.

---

The project tabs

Every project has nine tabs:

Tab	What it is for	Page
Builder	the questionnaire: pages, questions, codebook, logic, theme	The Builder
Distribute	publishing, environments, links, QR codes, invitations	Publishing and Environments
Data	the response database, insights, exports	Responses and the Data Tab
Flows	analysis flows	Analysis Flows
Live	the fieldwork monitor and live tiles	Live Monitoring
Reports	the reports your flows produced	Reports
History	every Save: diffs, restore, bundles, Methods	History and Versions
Files	uploads and run outputs	Files
Settings	name, citation, runtime, environments, secrets, deletion	Project Settings

Distribute shows the number of live environments and Flows the number of flows next to their names. On screens narrower than 1280 pixels the tabs after Reports move into a More menu; below 1024 pixels, the tabs after Flows.

---

Renaming and deleting

Both are in Settings and are for owners and admins only:

• Settings → General → Project name, then Save changes — 1 to 120 characters. The slug does not change. A member sees Save changes disabled, with the note "Only owners and admins can rename a project."
• Settings → Danger Zone → Delete project — you type the slug to confirm. Deleting removes the questionnaire, the flows, every Save, all deployments and every response, immediately and permanently; live survey links stop accepting answers. Download a research bundle and export your data first. A member sees Delete project disabled ("Only owners and admins can do this"), with the note "Only owners and admins can delete a project."

Details in Project Settings.

---

Practical conventions

• One study per project. Waves of the same tracker belong in one project, in separate environments or successive deployments, so the codebook and the history stay together.
• Name projects for humans (Employee Pulse 2026 Q1). A name in Cyrillic or Greek gets a readable transliterated address; a name written only in a script with no Latin spelling (Chinese, Japanese, Arabic, …) gets the address /project, so add a few Latin letters or digits if you want the address to say what the study is.
• Save early, Save often, with messages. Saves are cheap, and the message is what makes History readable six months later.
• Start from the example study once. It shows every part of the product working together on data you can safely break.

See also

• The Builder
• History and Versions
• Project Settings
• Question Bank, Templates and Library
• Key Concepts

---

← Organizations and Team · Studio contents · Project Settings →