Studio Tutorial End to End

Tutorial: A Study from Start to Finish

This tutorial follows one realistic study through every stage of Studio — from an empty project to a published report and a citable research bundle. It takes about two hours the first time. Each part ends with a checkpoint, so you can stop and come back.

The study. Customer Pulse 2026: a short satisfaction survey for the customers of a regional service, fielded in three regions, with a consent page, an age screener, an attention check, a Net Promoter Score, a satisfaction rating and an open comment for unhappy customers. The analysis cleans the data, weights it to the regional customer mix, tabulates satisfaction by region and publishes a report and a live dashboard.

Part 1  Plan                     Part 5  Build the analysis on simulated data
Part 2  Build the questionnaire  Part 6  Pilot, pre-register, launch
Part 3  Make it look right       Part 7  Monitor fieldwork
Part 4  Test it                  Part 8  Report, close, archive

Before you start. You need an organization on a plan that includes the features used here — the 30-day Pro trial covers all of them. Parts marked (Plus) need Plus or higher after the trial. See Plans, Trial and Billing.

A finished study to compare with. The example study (Projects → New project → Template → Example study) is a larger study built with the same tools and already filled with 729 sample responses. Its flow cleaning does what Part 5 builds — quality checks, raking, a table the other flows read, a live tile — with Response quality in drop mode rather than flag; tables adds a banner table, a tab book and coded open answers; and wellbeing_app reports a Net Promoter Score. Open it in another tab to see each step done on a study that has data — see The example study's flows.

Part 1 — Plan

Write down, before touching the Builder, what you need to be true of the data. For this study:

Decision	Choice
Who may answer	adults (18+) who agree to take part
Sample	300 completed interviews, roughly 100 per region
Key measures	satisfaction (5-point), NPS (0–10)
Quality	one attention check; flag, don't drop
Weighting	to the known customer mix: North 45 %, Center 30 %, South 25 %
Deliverables	a report with a weighted table and a chart, a live tile for the client, a research bundle

Two naming habits make everything downstream easier:

Name variables for analysis — consent, age, region, satisfaction, nps, attention, comment. Each answer is stored under its variable name: that is the data column, and the name conditions, piping and quota counting read.
Give each question the same Id as its variable. The Id is the question's handle in the Builder: the Logic map and validation messages name questions by it. Logic works whatever the Id is, but satisfaction is easier to read there than q7. The one thing Studio refuses is an Id that is another question's variable name. See Question Id and variable name.

Part 2 — Build the questionnaire

Create the project

Projects → New project. Name: Customer Pulse 2026.
Start from: Blank project → Create →.

The Builder opens on page1, which holds one placeholder question ("New question", a single choice with Option 1 / Option 2, variable q1). The first Save, #1, is valid.

Page 1 — consent

Select the page in the rail. In the Inspector (Page), set Title to About this survey and Name to consent_page. The new name takes effect when you leave the field or press Enter (Esc puts the old one back).
+ Question → Presets → Yes / No. Set the Question text to Do you agree to take part in this survey? and turn Required on.
Open Variable and change the variable name to consent (press Enter). Then open Advanced and set Id to consent too.
Select the placeholder question q1 and press Delete question (the trash icon in the Inspector's header). Its variable q1 leaves the codebook with it: the message at the bottom says "Removed variable q1 too", with Undo.

The Yes / No preset codes 1 = Yes, 0 = No and shows them as buttons.

Introductory text for respondents goes in the page's Body (hint "HTML, shown above the questions …"): The survey takes about five minutes. Your answers are anonymous. It is HTML, not Markdown — wrap paragraphs in <p>…</p> if you write more than one.

Page 2 — screener

+ Page → Content page. Title About you, name screener.
+ Question → Types → Number. Text How old are you?, Required on, Unit years. Variable name age, Advanced → Id age. In Variable, set Min 16 and Max 99.
+ Question → Types → Single choice. Text Where do you live?, Required on. Choices: 1 North, 2 Center, 3 South. Variable and Id region. Variable label Region of residence.

Page 3 — experience

+ Page → Content page. Title Your experience, name experience.
+ Question → Types → Likert scale. Text Overall, how satisfied are you with our service? Points 5, Left label Very dissatisfied, Right label Very satisfied. Variable and Id satisfaction.
+ Question → Presets → NPS (0–10). Keep the standard wording. Variable and Id nps.
+ Question → Presets → Attention check. It is a required single choice (…please choose "Rarely") already marked as an attention check with its acceptable answer. Variable and Id attention.
+ Question → Types → Open text. Text You gave us {answer:nps} out of 10. What should we do better? Turn Multi-line on. Variable and Id comment.

{answer:nps} is piping: the respondent sees the number they chose.

The ending pages

+ Page → Final page — the thank-you page for completed interviews. Edit its Title and Body in the Inspector.
+ Page → Screen-out page — named disqualification, with the text You do not qualify for this study. Edit it if you like.

Check the order in the rail: consent_page, screener, experience, final, disqualification. Drag pages to reorder if needed. Keep the screen-out page last, after the final page — it is reached only by the rules you add next.

Routing

Screen out non-consenters. Select consent_page. In Logic → Branch (next if) press + Rule. A draft rule opens with its condition editor already showing a comparison on the page's last question: Variable consent, operator =. Pick the value No (0), set the target (→) to disqualification, and press Add rule — it stays disabled until the comparison has a value ("Pick what the rule tests — it is added once the condition is complete.").

Screen out minors. Select screener and press + Rule. The draft opens on region, the page's last question: change Variable to age, the operator to < and the value to 18, target disqualification, Add rule.

A branch rule with an empty condition never fires — it is not an "otherwise". Studio does not add a draft rule until its condition is complete, but a rule can lose its condition later (Clear) or arrive without one from an import; then the Inspector shows "Add a condition — an empty rule never fires." under it, and the Logic map labels its arrow "no condition — never fires". Use Default next for "everyone else".

Ask for a comment only from detractors. Select the comment question. Logic → Show if → Add condition: nps ≤ 6 → Done.

Also screen out inattentive respondents? For this study, no — we flag them in the analysis instead. (The attention check's Also end the survey for respondents who fail option would add a branch rule from experience to a Screen-out page placed after the Final page — here disqualification, which already sits there — so that respondents who pass still finish on final.)

Check the routing

Open Logic map. The Pages lens should show consent_page → screener → experience → final, with rule arrows from consent_page and screener to disqualification. Switch to the Questions lens: the comment row carries a SHOW IF chip reading nps, and no arc is red (a red arc means a condition reads an answer given later).

Quotas

Open More ▾ → Quotas. In "or one for every value of", pick region, set the limit to 100 and press Add 3 cells.

Each cell closes once 100 completed interviews have its region; screen-outs and unfinished interviews do not count. A later respondent who picks a full region is stopped when they leave the screener page, on the quota-full screen: "Thank you for your interest" / "We have already reached our target sample for participants like you." Their interview is not submitted. Previews do not check quotas — publish to pilot with a small limit to see the screen. See When a cell is full.

Save

Press Save changes, message Questionnaire v1, Save. The badge should read ● valid #2 or ● warnings #2. If it says errors, open Validation and fix what it lists.

Checkpoint. Five pages, seven questions, three quota cells, a valid Save.

Part 3 — Make it look right

Open More ▾ → Theme.

Appearance — set a Primary color that matches your brand; read the contrast line under the colors (it turns red below the WCAG floor). Leave Progress on bar: published surveys show the indicator this setting chooses.
Branding — Institution Customer Insights Team. For a logo, paste a Logo URL from a stable public address (your website). Do not use a link from Files — those download links expire after a few minutes.
Respondent experience — fill Contact email, Privacy URL and the Ethics statement. They appear in the footer of every page. Set Estimated minutes to 5: respondents see "About 5 minutes" under the title of the first page (you can then drop the sentence about the length from the consent page's Body).
Wording — optionally rephrase the runtime's fixed texts: buttons, page labels, answering hints and error messages, saving and failure messages, the screen-out and quota-full screens. The notices a published survey shows when it is closed or paused stay in English.

Note. Completion screen has a Title and a Message, but they are shown only where a survey ends without its own ending text — on Submit of a question page, or on a Final page with no title (for the Title) or no body (for the Message). This study ends on its own Final page, whose Title and Body you already set, and the hints beside the two fields say that the Final page's own text is shown instead. See Completion screen.

Switch the canvas to Structure | Preview and check both Desktop and Mobile. Save (Theme).

Part 4 — Test it

Validation

Open Validation. The Engine section should report no errors (press Check now after edits). Read every warning; lint remarks such as an unused variable are fine if you understand them.

Walkthrough — both paths

Test → Walkthrough. Take the survey as a consenting 35-year-old from the North who is unhappy (NPS 4): the side panel should show comment evaluated as shown. Restart and answer No to consent: the panel should report the consent_page rule matching and routing you to disqualification. Restart once more as a 16-year-old.

Simulate

Test → Simulate: Responses 300, Seed 42, Simulate. Look at the preview rows and the codebook, then download SPSS and open it: are the variable and value labels right?

Simulation follows the questionnaire's logic the way a respondent meets it: page, block and question conditions, routing and screen-outs, the arm of Assign to a condition, Randomize pages and quotas — a simulated respondent who answers into a full region cell stops there. Option shuffles and custom JavaScript are not simulated. See Simulate.

Share preview for reviewers

Test → Share preview creates a public link (copied automatically) that a colleague or client can open without an account for 24 hours. Answers are not stored. Collect wording comments; apply them; Save.

Checkpoint. Both routes behave; the simulated SPSS file is labeled.

→ Testing Your Survey

Part 5 — Build the analysis on simulated data

Build the whole analysis now, before the first real respondent arrives.

Create the flow

Flows → New flow, Title Satisfaction, Open canvas. The canvas starts with a Responses source; delete it for now (select it, Delete).

Wire the pipeline

Drag these nodes from the palette and connect each node's data output to the next node's data input:

Simulated data ─▶ Response quality ─▶ Rake weights ─▶ Apply weight ─┬─▶ Banner table ┐
                        │                                             ├─▶ Bar chart ─┤
                        │                                             └─▶ Net Promoter Score
                        └──(table)───────────────────────────────────────────────────┤
                                                                                      ▼
                                                                          Report section ─▶ Save report
Apply weight ─▶ Live tile

Set the parameters:

Node	Parameter	Value
Simulated data	Respondents · Seed	`300` · `42`
Response quality	Attention checks	use Fill from the questionnaire — it fills `attention` with its acceptable answer
	Mode	`flag` (adds `quality_flags` and `quality_score`, keeps everyone)
Rake weights	Targets (variable → code → share)	`{"region": {"1": 0.45, "2": 0.30, "3": 0.25}}`
Apply weight	Weight column	`weight`
Banner table	Questions (down) · Breakdowns (across) · Significance letters	`satisfaction` · `region` · on
Bar chart	Variable · By	`satisfaction` · `region`
Net Promoter Score	0–10 item	`nps`
Report section	Heading · Text	Satisfaction by region · a sentence of context
	Captions	one per connected item
	Note	Base: all respondents, weighted by region.
Save report	Title · File name	Customer Pulse 2026 · `satisfaction` — already filled in from the flow's name; it is saved as `outputs/satisfaction.md`
Live tile	Kind · Label · Show	`number` · Respondents so far · `rows`

Connect the Banner table's table, the Bar chart's chart, the NPS table and the Response quality table (counts per quality flag) to the Report section's items input in the order they should appear.

Which outputs are weighted? After Apply weight, the node's Weight column help lists them: Frequencies, Crosstab (Fisher's exact test counts respondents), Group means (not N or the test), Descriptive statistics (not N, skewness or kurtosis), Banner table, Net Promoter Score, Regression, TURF, MaxDiff, Conjoint, Share of preference, Principal components, Scale reliability, Key drivers, Perceptual map (its chi-square test counts respondents), Price sensitivity, Correlation and Correlation matrix with Pearson, Bar chart, Heatmap with By or with Pearson, Likert chart, Proportion CI with Weighted checked, Trend and Tab book (Excel); a Result chart follows the result it draws. A Banner table is used here for its significance letters; a Crosstab (Rows satisfaction, Columns region, Percentages col) would also give weighted column percentages, with a chi-square test on the effective base instead of letters. The Bar chart here, with By region, shows weighted mean satisfaction per region (its axis reads "Weighted mean …"). Compare groups, Correlation and Correlation matrix with Spearman or Kendall, t-test, Paired tests, Factor analysis, Cluster (k-means), Box plot, Scatter plot, a Spearman or Kendall Heatmap, Response quality, Code open answers, Data check, and the counts of MaxDiff scores and Bands stay unweighted and say so in their output. See Making tables and tests use the weight.

Preview and check

Select Banner table and press Ctrl/Cmd + Enter (Run to here) — the weighted table appears in the inspector, with column percentages, counts and significance letters.
Press Check — the engine should report no errors. Fix anything it lists before saving: a flow saved with engine errors is stored without code, so it cannot run (its Run button is disabled and the flows table marks it errors) until you fix it and save again. The questionnaire and the project's other flows are not affected.
Switch to Report view to write the words around the outputs and use Preview report.

Save changes (Analysis v1), then Run. The run appears in Run history; the report appears on the Reports tab.

Make the tile live (Plus)

With no node selected, the inspector shows Flow settings. Check Live: recompute on new responses and Save. New responses will re-run the flow and refresh the tile.

Checkpoint. A saved, runnable flow and a report — all on simulated data.

→ Analysis Flows · Cleaning and Weighting Data · Node Reference

Part 6 — Pilot, pre-register, launch

Pilot

Distribute → Publish panel: Target pilot → Publish to pilot.
When the card shows ● Live, Copy the link and send it to three colleagues. Answer it yourself on a phone.
Check Data → responses: are the columns what you expected? Is comment empty for promoters?
Fix anything you find in the Builder, Save, and press the Republish #N chip on the pilot card.

Switch the analysis to real data

In the flow, drag in a Responses node, set Environment main and check Only completed responses. Connect its data output to Response quality — this replaces the connection from Simulated data — then delete Simulated data. Check, Save.

Only completed responses leaves out unfinished interviews only. Screen-outs — here the non-consenters and the minors — were submitted, so they stay in the flow, marked screened_out in __status. To analyze only the people who qualified, put a Filter rows node between Responses and Response quality with the condition consent = 1 and age ≥ 18.

Pre-register (optional)

History → open the latest Save → More ▾ → Pre-register. The Save is tagged as the registered questionnaire and analysis plan; later Methods drafts and bundle provenance report what changed since. For an external timestamp, also Deposit it to Zenodo (Part 8 shows how).

Launch

Distribute → Publish panel: Target main → Publish to main. Share the main link: email it, print the QR code, or paste the Embed snippet into your website.

New projects cap main at 1,200 completed interviews and pilot at 50; screen-outs and unfinished interviews do not count. On the Free plan the project also stops at 1,000 completed interviews, all environments together. When a cap is reached, a respondent who opens the link sees "Thank you for your interest — We have already reached our target sample for participants like you." at once; someone who was already answering sees it when they submit, and their interview is not stored as completed.

Alternative channels:

a sample provider — Panel Providers;
personal email invitations with reminders (Plus, after the first payment) — Email Invitations;
restricted access with codes, or One per browser to refuse a second interview from the same browser — Links, QR Codes, Embeds and Access Control.

Checkpoint. main is live; the analysis reads real responses.

Part 7 — Monitor fieldwork

Distribute tiles: Responses · main — the completed interviews against the cap (the tooltip gives all rows, partial and screened-out ones included) — Completion, Median duration (and speeders), Quality screen (share failing the attention check), Today.
The main card: one bar per quota cell (region=1 · 64/100). When a region reaches its target, the cell closes by itself: later respondents from that region end on the quota-full screen when they leave screener, and the other regions go on. To collect more from a region, raise its limit, Save and Republish #N.
Drop-off on the card: which page people abandon.
Live tab: project totals, a responses-per-day chart and the flow's tiles. Create public link gives the client a read-only page with just the tiles (Plus); Revoke it when the engagement ends.

→ Live Monitoring · Data Quality

Part 8 — Report, close, archive

Close the field

On the main card press Close → Close survey. Respondents now see "This survey is closed". (Reopen would rebuild the same Save into the same link.)

If you know the end date in advance, set it instead: the Closing date chip on the main card → pick the date → Save date. It applies at once, without a new Save or a rebuild. When the date passes, collection stops by itself, anyone who opens the link sees "This survey is closed", and the card reads ○ Closed with an Extend button in case you need a few more days — see Deadlines.

Final run and report

Flows → Satisfaction → Run. On Reports, open the report and read it. In the flow's Report view, the Look tab sets typeface, density, table style and page size (a4 for printing). Download HTML for the client or use Print / PDF.

Export the data

Data → responses → Export ▾ → SPSS gives a labeled .sav of the whole table: all environments, partials and screen-outs included — filter on survey_id, partial and __status (screened_out). The answers come in the questionnaire's order, followed by the fieldwork columns (duration_s, started_at, url_<name> for each link parameter, the captcha verdict and the tab-switch counts). Stata gives the same as a .dta.

Citation, bundle, deposit

Settings → General → Study & citation: title, authors (one per line: Name; ORCID; affiliation), license, keywords, abstract → Save study metadata.
History → open the final Save → More ▾ → Methods: a draft Methods section built from your documents; everything marked [...] needs you.
More ▾ → Download a bundle → With the responses so far: the study as a zip — questionnaire and flow code, documents, codebook, METHODS.md, CITATION.cff, PROVENANCE.md and the data (the responses, and the project tables and uploaded files the flows read). The survey link's parameters, such as panel ids, are included only where a flow reads one, and invitation tokens never are.
More ▾ → Deposit (needs a Zenodo token stored under Settings → Secrets, which an owner or admin adds): choose Zenodo, pick the token, uncheck "Use sandbox.zenodo.org" for a real DOI, decide on Publish immediately, and Deposit.

Done. The study, its analysis and its data are versioned, reproducible and citable.

→ Reports · History and Versions · Reproducibility: Code, Bundles and Citation

What you practiced

Skill	Where it is documented
Pages, presets, Ids and variables	The Builder, Question Types
Branch rules, show if, piping, Logic map	Logic and Branching
Quota cells and the quota-full screen	Quotas and Randomization
Theme and footer details	Theme and Branding
Validation, walkthrough, simulation, share preview	Testing Your Survey
Flows, quality flags, raking, reports, live tiles	Analysis Flows, Reports
Environments, pilot, launch, response caps, closing dates	Publishing and Environments
Pre-registration, bundles, deposits	History and Versions

To see the same skills on a larger study, with six flows and their reports already built, start a project from the example study.

Studio Tutorial End to End

Tutorial: A Study from Start to Finish

Part 1 — Plan

Part 2 — Build the questionnaire

Create the project

Page 1 — consent

Page 2 — screener

Page 3 — experience

The ending pages

Routing

Check the routing

Quotas

Save

Part 3 — Make it look right

Part 4 — Test it

Validation

Walkthrough — both paths

Simulate

Share preview for reviewers

Part 5 — Build the analysis on simulated data

Create the flow

Wire the pipeline

Preview and check

Make the tile live (Plus)

Part 6 — Pilot, pre-register, launch

Pilot

Switch the analysis to real data

Pre-register (optional)

Launch

Part 7 — Monitor fieldwork

Part 8 — Report, close, archive

Close the field

Final run and report

Export the data

Citation, bundle, deposit

What you practiced

See also

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!