Decision Pro
Public API
Read-only REST API with your own keys and scopes: retrieve surveys, insights and decision summaries and bring them into your data warehouse or reporting.
For professionals
User research software that researchers can verify.
If research is your job, you have good reasons to distrust an AI analysis. That is why in usabiq every statement can be traced back to the individual response, every metric can be recalculated and the AI can be switched off completely. Language models only detect themes in free text — scoring, executive summary and segment comparison run without them.
- Methods
- 7
- AI switch
- per workspace
- Export
- CSV · API
- Traceability
- Every AI statement leads back to the raw response Clusters can be expanded, renamed, merged and split again; every assignment can be overridden individually. What the AI suggests is a draft — the analysis you stand behind in the end is yours.
- Where AI works
- Theme detection on free text, everything else is deterministic Automatic theme detection runs on JTBD Switch, JTBD Story and the free-form questionnaire — where free text is the material. ODI, Needs, SUS and UEQ+ are scored and analyzed without a language model. The executive summary and the segment comparison are rule-based and statistical, not generative, too. With very large volumes of free text the model reads a sample balanced across sources; shares are calculated against the full number of responses, and the unassigned remainder stays visible.
- Scoring
- Calculated by the book, with transparent intermediate values ODI opportunity score, UEQ+ on up to 15 scales with confidence interval, SUS with standard scoring from 0 to 100. Importance, satisfaction and all scale means can be viewed and exported individually — not just the end result.
- Statistics
- Segment comparisons with confidence instead of colorful bars Rating deltas and frequency differences are reported with statistical confidence. If a subgroup is too small for a statement, usabiq says so instead of showing a number nobody can defend.
- Your own instruments
- A free-form questionnaire when no method fits Likert scales, sections, single and multiple choice and free text can be combined freely. You don’t have to adopt a framework that doesn’t fit your research question — the AI analysis runs on open answers regardless of the method.
- Control over the AI
- One switch disables all AI processing The workspace switch overrides every plan — usabiq then becomes a pure survey and analysis platform. Alternatively, choose a European provider (Mistral in France, T-Systems in Germany) or add your own keys on Enterprise. Personal fields are removed before every call anyway.
- Agencies
- Clients kept separate, results under your name A separate workspace per client, results as a password-protected link with an expiry date — clients need no usabiq account and don’t count as a seat. Your own branding on shared results and whitelabel with your own domain are available in the agency plans — they are not in the price table; we agree on them in a conversation. The audit log for proof towards your client is included from Decision Pro.
- Raw data out
- Export without detours, if you want to run your own numbers CSV export of responses and metrics from Research Core, plus a read-only REST API for surveys, insights and decision summaries from Decision Pro. Your data stays yours — outside usabiq, too.
Want research to run as part of existing processes — backlog, ticket, data warehouse? You’ll find the API, webhooks and ready-made integrations in the next section.
Research that flows into your business processes.
Results don’t stay inside usabiq. From Decision Pro you pull insights via the API or have events pushed into your systems — and they land where decisions are made anyway: in the backlog, in the ticket, in the channel.
Decision Pro
Webhooks
Signed events sent to your endpoint as soon as something happens:
survey.completed, insight.generated,
decision.ready. No polling needed.
Enterprise
Slack · Jira · Productboard
Ready-made integrations without writing code: a finished decision summary creates Jira issues per top priority, a Productboard note or a message in your team channel.
The list keeps growing. Missing an integration you need? We also build integrations on request — write to us.
Other tools show charts. usabiq tells you what they mean.
The analysis is not a second tool you open afterwards — it runs on the same responses, in the same platform, with the same segments.
Text clustering instead of manual tagging
The AI reads every open answer, groups it by theme, adds sentiment and shows the demographic distribution per cluster. No codebook, no late nights in spreadsheets.
Cluster with supporting quote
I had to ask three colleagues how it works first.Cluster: Getting started takes too long · 84 responses
The analysis saves me time every week.Cluster: Analysis saves time · 57 responses
JTBD patterns across all responses
Top jobs, recurring pains, anxieties and habits are extracted automatically from the JTBD responses — with verbatim quotes as evidence, not just a score.
Extracted forces · JTBD Switch
| Force | Mentions | Strength |
|---|---|---|
| Push — analysis too slow | 72 | high |
| Pull — methods already built in | 58 | high |
| Anxiety — participant data privacy | 31 | medium |
| Habit — existing Excel templates | 24 | medium |
Compare segments, don’t just filter
How did product managers answer compared with developers? Rating deltas, frequency differences and statistical confidence side by side.
Segment comparison · UEQ+ efficiency
| Segment | Avg. rating | Delta |
|---|---|---|
| Product management | 1.8 | +0.6 |
| Development | 1.2 | −0.4 |
| Design | 1.6 | +0.2 |
Privacy first, from day one.
Participants trust you with their feedback. usabiq is built so that this trust remains justified.
- Built for GDPR Designed to European data protection standards from day one.
- PII encryption Sensitive participant data encrypted field by field with XChaCha20-Poly1305, authenticated, with a separate key per tenant.
- Privacy notices Generated automatically for each survey, and editable.
- Verifiable consent Participant consent is logged and can be verified.
- No use for other purposes Your data is processed only to provide the service — not sold, not used for advertising, not used for AI training. We name the processors we use on request.
- Audit logging A complete trail of actions for compliance teams.
How the methods differ in their analysis is shown in the methods comparison. In-depth articles on trusting AI analysis and on GDPR-compliant surveys are available on the blog (in German).
Frequently asked questions about AI, data and integration
How exactly does the Decision Engine work?
The AI rates every insight on five criteria — impact, frequency, criticality, reliability and segment relevance. Each insight then gets a clear label (problem, opportunity, quick win, risk, segment topic, research need) and is sorted into a prioritized list: top 3 problems, top 3 opportunities, all quick wins.
Which AI does usabiq use?
usabiq supports four providers — Anthropic Claude and OpenAI (USA) as well as Mistral AI (France) and T-Systems LLMHub (Germany) — and automatically fails over to the next one if a provider is down. You don’t have to configure anything. Before each call, structured profile fields are removed and email addresses and phone numbers are masked in free text; for other personal details that participants type in freely, the survey owner remains responsible. Enterprise customers can add their own API keys (BYOK) and choose their provider. If you don’t want AI at all, switch it off for the whole workspace — the switch overrides every plan.
Is my data used to train AI models?
No. Your data is never used to train AI models — neither by us nor by the AI providers. Structured profile fields are removed before any AI processing, and email addresses and phone numbers are additionally masked in free text. The free-text answers themselves go to the provider for analysis — what participants type there cannot be fully detected technically. Every customer gets their own isolated environment.
Where is my data hosted?
Your survey and participant data is processed solely to provide the service — not sold and not used for advertising. Operations are being moved to a data center in Germany; we will tell you the current location and the processors involved on request. Audit log and automatic retention control are included. European providers are available for text analysis (Mistral AI in France, T-Systems LLMHub in Germany); the current default provider is Anthropic in the USA, and European providers can be selected on request. The vector computation for semantic search and “Ask usabiq” runs exclusively through a US provider — there is no European alternative for it yet.
Is there an API for integrations?
Yes, from Decision Pro. You create your own API keys for the read-only REST API (surveys, insights, decision summaries) and subscribe to signed webhooks for survey.completed, insight.generated and decision.ready. Ready-made integrations with Slack, Jira and Productboard are part of Enterprise. More integrations are added continuously, and we also build them on request.
Ready for decisions based on data?
Start with one method, one question and one shared link — the analysis doesn’t wait for next quarter.
14 days of Decision Pro, no credit card required · then free on Free Discovery for good