UserTesting Alternatives: Lookback, Maze & Userlytics
Compare Lookback, Maze and Userlytics as UserTesting alternatives. Test recordings, outcome codes and exports, including help and incomplete attempts.
Lookback, Maze and Userlytics are UserTesting alternatives worth comparing for live session review, structured prototype tasks and screened prototype-link studies. The right choice depends on what your team needs to see and recover later. Imagine a subscription app checking its promise that cancellation is easy.
This comparison is published by Sapien.
About these examples: the cancellation session and separate delivery-address records below are illustrative. They show how a research workflow can change the apparent result; they are not vendor-performance measurements. Use a test account or prototype that cannot alter a real subscription.
One completed cancellation can still reveal a problem
Ask a relevant user to stop the next renewal. Don't name the control they need to find. Decide in advance what counts as success: reaching settings or opening a cancellation page may leave the renewal active. Record how the person reads the warning as well as what they eventually do.
Field | Session record |
|---|---|
Participant / stimulus | P01 / v1 |
Task | Your plan renews next month. Find how to stop the renewal. |
Evidence window | 02:10–03:05, a 55-second excerpt |
Observed account | Control found after a moderator hint; the hint’s exact timestamp is not supplied. |
Outcome code | Assisted success, retained in the eligible task denominator |
Design action | Inspect the label or route where help was needed, then test the changed interface. |
The clip is 55 seconds of a longer task, and the moderator had to help. Count it as an assisted success. The record lacks the hint's exact timestamp and the full task's start and end, so it cannot answer every timing question. Its useful lesson is where the label or route needs attention.
Now ask a colleague to read the report. If it says 'navigation failed,' can they return to the session, find the help and recover the outcome? A different code can be corrected. A missing recording cannot.
What the three alternatives put within reach
Alternative | Evidence and scope to verify |
|---|---|
Live stakeholder observation, retrievable recordings, transcripts and session exports. Do not assume Maze-identical automated path scoring. Confirm recruitment through User Interviews or your audience. | |
Repeated prototype tasks with completion, misclick and mission-path metrics. Pilot the success definition and stimulus fidelity; a score alone may not explain confusion. | |
Prototype-URL tasks with demographic filters and behavioral screeners. Verify permissions, mobile behavior, eligible participants and export fields; broad compatibility is not a tested migration. |
The official feature and prototype pages behind this table were checked on October 6, 2026; no accounts were tested for the comparison. UserTesting itself advertises video and clickstream evidence. List the parts of that workflow your team actually uses before replacing it with something narrower.
Lookback deserves attention when colleagues need to watch together, ask follow-up questions and retrieve recordings or transcripts. Maze is relevant when repeatable prototype missions and path metrics drive the work. Userlytics may fit prototype-link tasks requiring the right demographic filters and behavioral screeners. Pilot the actual device, audience, permissions and export.
A pilot that reaches the analyst
Save the task wording, prototype version and device. Set the start, end and assistance rules, then recruit people with relevant subscription experience. Keep unassisted completion, assisted completion and incomplete attempts distinct. The analyst should be able to follow each outcome back to what happened.
Requirement | Evidence and acceptance judgment |
|---|---|
Cancellation task | Example evidence: Test participant P01, stimulus v1, recording 02:10–03:05; found control after moderator hint Acceptance judgment: Assisted success; keep in denominator |
Traceability | Example evidence: Report says “navigation failed,” with the excerpt window and reported moderator help preserved; the hint’s exact timestamp is unavailable Acceptance judgment: Revise the code to assisted completion; do not count as unassisted success |
Export | Example evidence: Example export retains participant ID, task, stimulus and assistance flag Acceptance judgment: Accept this evidence structure; still demonstrate actual vendor export |
Recruitment | Example evidence: Screener asks whether participant manages a subscription Acceptance judgment: Check relevant participants are available within quote |
Include a broken path as well as a success. A platform can collect a polished success clip while losing the incomplete attempt that explains the design issue. Give unusable sessions an explicit status so someone else can reconstruct the base.
Ask a researcher who did not run the session to recover the result from the proposed output. If an assistance flag is absent, find out how much manual coding is required. Price that review work, recruitment and exports as part of the replacement. A dashboard label is only the beginning.
Maze alternatives: the denominator can move without the design improving
The same check matters when replacing Maze. Recordings, live probing, structured task outcomes and automated path measures serve different purposes. First reproduce the measure on a fixed prototype; test a redesigned interface after the migration comparison works.
Here is the separate delivery-address task, with ten eligible attempts:
Outcome | Attempts |
|---|---|
Unassisted success | 5 |
Assisted completion | 2 |
Wrong end state | 1 |
Stopped | 2 |
Counting assisted completions as successes and dropping the stopped tasks gives 7/8, or 87.5%. Counting only unassisted success across everyone eligible gives 5/10, or 50%. The interface and attempts are unchanged. The reporting rule creates the gap.
Retain the successful end state, help policy, incomplete attempts and timing rules. Give a second researcher the proposed export and ask them to recover the four outcomes in the table. That is a fair migration check. Only then should a new design get credit for changing the result.
The promise and the interface need their own answers
A person may expect cancellation to be easy yet struggle to find it. Sapien's managed research service can compare modeled reactions to the subscription promise or wording; a human task test shows how people use the cancellation interface. The message-testing guide and concept testing versus usability testing explain how those questions fit together.
For recurring work, put the study's recruitment, evidence, outputs, validation, total cost and division of work into the research brief template. Ask the provider which part it cannot answer. Our System Usability Scale guide addresses the broader rating question; a scale score does not replace the cancellation episode.
For other human-research workflows, compare dscout. Use the synthetic and human respondents guide when deciding which source the next question needs.
Keep the cancellation pilot focused on behavior. Choose the workflow in which a colleague can rewatch the task, understand the help and reach the same conclusion you did.
