Field faster. Script less.
Deliver clean data.
Fieldbase reads your questionnaire and writes the survey from it. Your programmer reviews the code instead of hand-coding it, and every language launches from the same build. Built for research agencies and fieldwork teams.
- qid: Intro type: info title: Welcome text: | Thank you for participating in this research about streaming video services. The survey takes about 12 minutes.- qid: S1 type: single text: In which country do you currently live? instruction: <i>(Single response)</i> required: True order: fixed skin: radio options: 1: United States 2: United Kingdom 3: Canada 4: Australia# Punch hMarket immediately after S1 — carries the# country label for piping and quota.- script: | if S1.value is not None: hMarket.set(S1.value)- qid: hMarket type: single visibility: hide title: Market (from S1) — used for pipingBuilt for the teams that field.
Pick your team for the reasons that matter to you, or keep scrolling for the workflow all four share.
More studies per programmer, without the next hire.
- 60% less work per study. On average. The AI drafts from the client’s questionnaire and your programmer reviews it.
- Late client changes, on the record. Each edit arrives as a tracked change you accept or reject, logged in the project thread: who asked, who accepted, when.
- Tables without a tab house. Crosstabs with banners, significance letters and the Excel layout clients already read.
- Sell conjoint in-house. CBC and MaxDiff design, fielding and simulators in the same project, on the Team plan.
Win the bid on turnaround. Keep the PM in control.
- Quote programming in hours. First preview 5x faster from the client’s spec, so your turnaround is the one that wins.
- Supplier-ready from day one. Respondent IDs pass through, and every exit redirects back to the supplier: complete, terminate or quota-full.
- Exact counts, fair exits. Completes are counted once under a lock. Full cells screen out at the door, not after 25 minutes.
- Launch markets independently. Each language goes Live, Paused or Closed on its own, so one late translation holds nobody up.
The AI drafts the code. You make the calls.
- An IDE for surveys. Code, structure and live preview side by side. Readable YAML with Python-like logic, no drag-and-drop.
- Every AI change is a diff. Accept, refine or skip. Nothing is rewritten behind your back.
- No ceiling. Custom HTML, CSS and JavaScript at question, page or survey scope when the client wants bespoke.
- Free kit, freelance plan. Learn on the free developer kit. Solo is $59 a month for your own clients.
- qid: Q4 type: multi options: 1: Netflix 2: Disney+ 3: Prime Video + 99: + label: None of these + exclusive: True + anchor: True
Agency-grade research, without the agency or the enterprise contract.
- No programmer required. Upload the questionnaire, then change it by asking in plain English.
- Conjoint and MaxDiff, built in. Run choice studies, then play what-if in the market simulator.
- Share results, not files. Send stakeholders a report link, and revoke it whenever you need to.
- Passes procurement. Published per-seat pricing, ISO 27001, and your data never trains AI models.
Upload a spec.
Get a working survey.
Drop the Word doc or paste the questionnaire. Fieldbase reads the question text, options, programming notes, and routing instructions — and writes the survey code for you to review.
- S1single
- {}script
- hMarketsingle
- S2single
- {}script
- S3single
- Q1multi
- Q2multi
- Q3gridSingle
- Q4cbc
- qid: S1 type: single text: In which country do you currently live? required: True options: 1: United States 2: United Kingdom 3: Canada 4: Australia
Describe the change. The AI writes it.
You accept, refine, or skip.
Scope with @, describe what you want. The AI proposes a diff — lint-checked and in your DSL — with Accept, Refine, or Reject buttons. Every change is logged in a thread the whole team can trace.
How AI assist works →- qid: S1 type: single text: In which country do you currently live? required: True options: 1: United States 2: United Kingdom 3: Canada 4: Australia# Market quota check — fires on S1 selection- script: | if hMarket.value is not None: cell = hMarket.value if not quota("qtMarket", cell).isOpen: endSurvey("quotafull")- qid: S2 type: number text: What is your age? min: 13 max: 99 required: True# Age gate — out-of-scope respondents terminated- script: | if S2.value >= 65: if S2.value >= 65: if S2.value < 18 or S2.value >= 65: endSurvey("terminate")- qid: Q1 type: multi text: Which of these brands have you used in the past 3 months? min: 2 options: brandList- qid: Q2 type: gridSingle text: How satisfied are you with each of these features? rows: brandListSee your results instantly.
Export clean data.
Auto-generated topline reports with cross-tabs and filters. Deterministic column naming across every export — no cleanup, no surprises between waves.
| Total | Dell | Apple | HP | Lenovo | |
|---|---|---|---|---|---|
| Quality | 3.8 | 3.4 | 4.5 | 3.2 | 3.6 |
| Value | 3.2 | 3.6 | 2.1 | 3.8 | 3.7 |
| Ease of use | 3.9 | 3.5 | 4.6 | 3.3 | 3.4 |
| Availability | 3.7 | 4.2 | 3.8 | 3.9 | 2.8 |
Built for how modern
survey work actually happens.
Each capability is designed around the way survey programmers actually work today — fast, code-first, AI-augmented, with the research methods and output quality serious projects need.
Code-first authoring
Clean, readable syntax for every question type and every routing rule. Autocomplete and live validation from a single contract — every rule defined once, enforced everywhere.
single · multi · grid · cbc · maxdiff · cardSort Scripting & routing →Conjoint and MaxDiff included
CBC with three response modes — choice, rating, ranking. Upload a design file, reference it in a single line. Utility-ready exports drop straight into your analysis tool.
design: project://laptop_cbc_v2 Choice modeling & CBC →Exact quota counts
Every completion counted once under a database lock, even under heavy concurrent traffic. Full cells screen respondents out at the screener, not after 25 minutes, and targets edit live.
quota("AgeGender", 1).isFull() Quota engine →Live respondent preview
Every change updates the preview as you type. Branching, loops, piped values, hidden logic — see exactly what a respondent sees, without deploying. Test routing in seconds.
Survey simulator →Analysis-ready exports
Deterministic column naming — every export column is mechanically derived from the question structure. No surprises, no manual mapping, no cleanup before analysis.
QBudgetr1c1 · QBudgetr1c2 · QBudgetr2c1 Export formats & datamap →Theme framework
22 professional themes with seven style variants. Every visual property is a token — change how a survey looks without touching what it does. Respondent-facing UI that holds up to client review.
Theme framework →Every capability, in depth.
The platform's real surface — question types, scripting, design assets, simulation — each with its own page you can read, share, or send to a stakeholder.
15+ types — single, multi, grids, cardSort, CBC, MaxDiff, lookup, rank. 60 skin variants for different respondent contexts.
Explore question types →
Python-like - script blocks. Conditional routing,
quota gates, computed hidden values, screenouts — real logic.
Upload CBC / MaxDiff design files, reference in one line, and verify precision before fielding with Design Health + Precision scorecards.
Explore choice modeling →Fake respondents against the real quota engine. Catch routing gaps, load-test quotas, verify CBC and MaxDiff coverage.
Explore the simulator →Prompt. Author.
In your language.
Prompt the AI in the language your team thinks in — French, Japanese, Spanish, German, Portuguese. The AI responds in-language; the generated content stays in that language too. Twelve languages supported today, more as teams ask.
qid, type, options) and
identifiers stay English, so any reviewer on your team can read
the YAML regardless of the content language.
Built for the way research data has to be handled.
Your data isn't used to train AI
Survey content, respondent answers, and the code you author are never added to AI training data. Your work — and your clients' data — stays your own.
ISO 27001 certified, GDPR aligned
Certified under ISO 27001 and aligned with GDPR through our parent company, Inginit. Data is encrypted in transit and at rest, with tenant isolation by project. Full details on the Fieldbase trust center.
You own and control your data
Export the full project — code, design files, responses — at any time. Delete on request, including from backups within standard retention windows. Any operational metrics we keep for product improvement are anonymised aggregates that can't be traced back to you.
Hosted in US (San Francisco), with EU (Frankfurt) coming soon. Fieldbase is built by Inginit. For certifications, sub-processors, retention windows, and to request a DPA, see the Fieldbase trust center.
8 to 12 agencies, by invitation.
Run your next three projects on Fieldbase free — up to 5,000 completes over 90 days, hosted in the US (EU coming soon). We programme your first two questionnaires with you; you programme the third with the AI. At day 60 we talk about staying, with a 24-month price lock.
Apply Join the Launch CohortOr email [email protected] · Replies within one business day.