Les outils ux research les plus adaptés dépendent du moment du projet : Maze ou Lyssna pour tester une arborescence et un prototype, Lookback pour conduire des entretiens, Typeform pour les enquêtes, puis Microsoft Clarity ou Contentsquare/Hotjar pour observer un site publié. En 2026, une première combinaison peut coûter 0 $ par mois, hors recrutement et indemnisation des participants.
Which UX research tools should you choose based on the project stage?
A UX research tool is software that collects data on users' behaviors, expectations, or pain points. In 2026, the right choice usually follows four steps: structure the information, validate a prototype, conduct qualitative interviews, and then observe usage after launch.
Before detailed design, Maze and Lyssna make it possible to run card sorting, tree testing, first-click tests, and questionnaires. Card sorting asks participants to categorize content; tree testing checks whether they can find information in a menu with no visual design.
For a clickable prototype, Maze makes unmoderated testing easy: each participant completes a scenario on their own, at their own pace. The free plan allows one study per month and five seats in 2026. A 30-day trial, documented by Maze in October 2026, allows up to three moderated or unmoderated studies and five members, with some premium features excluded.
Interviews require a different kind of tool. Lookback is designed for moderated sessions, with screen capture, recording, and stakeholder observation. The researcher can ask why someone hesitates, which an automated test cannot do. To place these methods in a broader approach, the guide dedicated to UX design and its principles covers the useful basics.
How much do UX research tools cost in 2026?
Software costs start at $0 per month in 2026 with Microsoft Clarity and Maze's free plan. A mid-range setup starts at $138 per month, excluding any taxes, by combining Maze Starter at $99 per month and Contentsquare/Hotjar Growth at $39 per month with annual billing.
| Stage | Tool | Main features | Published price in 2026 |
|---|---|---|---|
| Architecture and prototype | Maze | Prototype testing, card sorting, tree testing, first click, surveys | Free; Starter at $99/month, excluding any taxes |
| Architecture and prototype | Lyssna | Prototype testing, card sorting, tree testing, first click, surveys | Free entry-level plan |
| Qualitative interviews | Lookback | Moderated tests, capture, recording, observation | Variable pricing; recruitment and participant incentives extra |
| Surveys | Typeform | Structured questionnaires | Around $25/month according to published comparisons in 2026, excluding any taxes |
| Live site | Microsoft Clarity | Heatmaps and session replay | Free, with no traffic limit announced |
| Live site | Contentsquare/Hotjar | Heatmaps, replay, funnels, surveys | Free up to 200,000 monthly sessions; Growth starts at $39/month with annual payment, excluding any taxes |
The listed price is only part of the budget, however. Enterprise plans from Maze, Contentsquare/Hotjar, and UserTesting are quote-based. They may add study volume, per-participant fees, panel targeting, recruitment, and user incentives.
Lookback stated in 2026 that it charges a 6% management fee when incentives are distributed through its platform. Maze said it had a global panel of more than 7 million participants for its Enterprise offering; Lyssna announced more than 690,000 active participants in 124 countries. These figures come from the vendors and do not, on their own, measure the quality of recruitment.
On a small budget, it is better to have two tools that are actually used than a full platform left without a protocol. Microsoft Clarity for the live site and free Maze for prototypes and surveys provide a no-subscription baseline. Time for preparation, analysis, and reporting still needs to be funded, however.
Should you prioritize user tests or heatmaps?
User tests explain why someone runs into a difficulty, while heatmaps show where multiple visitors click, scroll, or pause. To decide before development, prototype tests are more useful. After launch, heatmaps and session replay are mainly used to identify recurring issues.
A heatmap aggregates interactions on a page. Session replay reconstructs a visitor's navigation as a recording. These tools show an ignored button or an abandonment, but they do not prove the user's intent.
The classic mistake is to interpret mouse movement as an opinion. A lightly clicked area may be useless, misunderstood, or simply visible too late. A short interview often makes it possible to decide. On the projects we run, we mostly see good results when behavioral data is used to form a hypothesis, then verified with users.
Microsoft says that in 2026, Clarity powers more than 2 million sites and apps. Its free availability does not make analysis automatic, though: without a clear goal, dozens of recordings take time and produce few decisions. You need to start with a question, such as form abandonment or confusion about pricing.
How do you organize a useful user test on a small budget?
A useful user test first and foremost depends on a specific scenario, participants who match the target audience, and a decision to make. UX research tools reduce data collection costs, but they do not fix a vague question or poor recruitment. In 2026, a simple study can be launched without a software subscription.
The protocol can stay short. The goal is not to ask whether the interface is appealing, but to verify whether a task is understood and achievable. Honestly, recording twenty journeys is of little value if no one has defined what will trigger a change.
A lean approach follows five steps:
- Frame a single decision to inform, such as keeping or moving a section.
- Recruit people similar to future users, without revealing the expected answer.
- Write a realistic scenario, with a task and an observable endpoint.
- Launch a pilot test to identify ambiguous instructions.
- Classify findings by frequency, severity, and cost of correction.
For a survey, Typeform and Maze collect structured responses. In September 2026, Typeform recommended collecting neither names nor contact details nor IP addresses when a questionnaire is presented as anonymous. The promise of anonymity must correspond to the data actually recorded.
From the agency side, the instinct is to test the costly choices as early as possible: navigation, understanding of the offer, and conversion path. Modifying a tree structure in Maze or Lyssna is light work; redoing development and content after publication is much less so. This validation logic also fits within a decision driven by interpretable data, rather than by an accumulation of metrics.
Are session recordings compliant with the GDPR?
Session recordings can comply with the General Data Protection Regulation, but simply installing a tool is not enough. Between 2020 and 2026, the CNIL classifies heatmaps and session replay as advanced analytics functions, generally subject to prior information, consent, and an opt-out mechanism.
L’exemption applicable à certains outils de mesure d’audience reste étroite. Un rejeu détaillé destiné à améliorer l’expérience utilisateur ne doit donc pas être présumé exempté. Selon les lignes directrices du Comité européen de la protection des données en 2026, un consentement valide doit être libre, spécifique, éclairé et univoque ; la formule générique « améliorer l’expérience utilisateur » n’est pas assez précise.
To limit risks, the setup must specify each purpose, reduce the data collected, mask sensitive fields, and allow consent to be withdrawn. The draft recommendation published by the CNIL in 2026 specifically proposes distinguishing technical error detection from UX improvement. A vendor's "GDPR-compliant" claim does not transfer the controller's obligations.
In 2026, Microsoft Clarity requires a consent signal to offer full replay and continuous journeys to users in the European Economic Area, the United Kingdom, and Switzerland. Without consent, certain features and journey continuity are limited. Standard recordings remain available for 30 days; Microsoft then retains 1 % of recordings, or 10 per day if that number is higher, for up to nine months.
The banner configuration, masking of forms, and documentation of purposes must be checked before any collection. More broadly, the issues tie back to the basic rules of the GDPR and choosing a legal basis, even when the tool does not use artificial intelligence.
Framing the methods, recruitment, and compliance before comparing subscriptions avoids most unnecessary spending. An outside perspective can especially help connect each observation to a design decision rather than producing yet another dashboard.
UX research tools FAQ
What free tool should you use to test a prototype?
Maze offers in 2026 a free Forfait limited to one study per month and five seats, with prototype tests and surveys. Lyssna also offers a free entry-level plan for architecture and prototype tests.
How many users should be recruited for an UX test?
The number of participants depends on the method, the diversity of the profiles, and the risk associated with the decision. A small qualitative test is used to identify difficulties, while a quantitative comparison requires more responses and a rigorously constructed sample.
Is Microsoft Clarity really free?
Microsoft Clarity is free with no announced traffic limit in 2026. The actual cost nevertheless includes collecting consent, configuration, analyzing recordings, and the time needed to turn observations into corrections.
Can a mobile app be tested before development?
A mobile app can be tested from a clickable prototype before development. Maze or Lyssna make it possible to assess screen comprehension, first-click behavior, and task flow without immediately funding the full code.