Analyze your results with the UserTesting MCP (Open beta)

On this page:

• What you can read from your results
• Before you read results
• Read the AI summary for a completed test
• List the sessions in a test
• Read the transcript for a session
• Ask for a readout across sessions
• What the UserTesting MCP does not return
• FAQ
• Related content

What you can read from your results

The UserTesting MCP is a connector, built on the Model Context Protocol, that lets your AI client work with your UserTesting account. This article covers reading results. To connect it, see Connect the UserTesting MCP to your AI client (Open beta). To create a test, see Create and launch a test with the UserTesting MCP (Open beta).

Every transcript comes back with a video link and timestamps, so a quote your AI client gives you can be traced back to the session it came from.

Note: The UserTesting MCP reads results from tests run in the new experience only. Sessions and transcripts from Classic tests are not returned. 

Before you read results

Reading results through the UserTesting MCP requires four things.

• A connected AI client. See Connect the UserTesting MCP to your AI client (Open beta).
• An account with the UserTesting MCP enabled, and Generative AI features on. Both are checked when your AI client lists your accounts. An account that has opted out of Generative AI features, or that has not been enabled for the open beta, is not returned — your client reports No Accounts with Generative AI and MCP access enabled. To request access, contact Alison Held at aheld@usertesting.com.
• READ permission on the test, in that workspace. The UserTesting MCP returns results only from tests you can already open in UserTesting. You do not need WRITE or ADMIN permission to read results.
• A test run in the new experience. Sessions and transcripts are not returned for Classic tests.

Read the AI summary for a completed test

The UserTesting MCP returns the test-level AI summary for a launched test. This is the fastest way to get the headline findings before you go into individual sessions.

To read the AI summary:

1. Wait until the test status is Complete. 
2. Ask your AI client for the AI summary of that test. For example, “Pull the AI summary for my checkout redesign test”.

The summary includes the research goal, an overview, the participant count, timestamps, and a list of key findings. Each finding has a title, a short summary, and a label:

Need · Pain point · Opportunity · Expectation · Process · Ease of use · Language · Trust · Satisfaction · Suggestion · Confusion

Every finding also links back to the specific task in your UserTesting results that produced it, so you can check a finding against the underlying session data. Open those links before acting on a finding.

If the summary is not ready yet, the client reports “AI summary not ready” for that test and tells you to try again once participants have completed it.

Good to know

Test-level AI summaries cannot be edited or regenerated through the UserTesting MCP.

List the sessions in a test

The UserTesting MCP lists the sessions in a test you have READ permission for. This is the step before pulling a transcript — it gives your AI client the session identifiers it needs.

To list sessions:

1. Ask your AI client to list the sessions in a test. For example, “List the sessions in my checkout redesign test.”
2. The AI client returns one row per session.

Each session row contains:

Two things to expect from the list:

• Sessions are not returned in participant-number order. Ask your client to sort them if order matters.
• Some sessions have no participant number. Sessions with no real participant behind them - an uploaded video, for example - return a blank participant number rather than a placeholder.

Read the transcript for a session

The UserTesting MCP returns the full, verbatim transcript for one session at a time. Pick the session from the list first or pull all session transcripts from a test.

To read a transcript:

1. Ask your AI client to list the sessions in the test.
2. Ask it to read the transcript for the session you want by participant number, or by asking it to work through the list.

The transcript comes back as timestamped WebVTT: the participant's words, verbatim, with a start and end time on every line. Alongside it, the client receives the session's video link, so it can cite a quote at the moment it was spoken.

Transcripts are pulled one session at a time. A test with many sessions takes your AI client several steps to work through. Ask for the sessions you need rather than the whole test when you only need a few.

Sentiment is available on request. Sentiment is not returned by default. Ask for it explicitly if you would like it included. For example, “Include sentiment in that transcript.” 

Good to know

The transcript is returned verbatim and unsummarized. If you want a summary, ask your AI client for one but the underlying transcript it worked from is the participant's actual words.

Ask for a readout across sessions

Because UserTesting MCP transcripts arrive with a video link and timestamps, you can ask your AI client for analysis that stays traceable back to the source session. Three patterns available today:

Pull quotes for a readout. Read the sessions in my checkout redesign test and give me the five strongest verbatim quotes about checkout friction, each with a link to that moment in the video.

Draft a report. Summarize the top usability issues across all sessions in my checkout redesign test, with a supporting quote and timestamp for each.

Move evidence into your own systems. Ask your client to pull transcripts into the research repository, wiki, or analysis pipeline your team already runs.

Tip: Name the test in your prompt. Your AI client searches tests by title, so an exact title finds the right test in one step. See Create and launch a test with the UserTesting MCP (Open beta) for how test search works.

Check any finding against the video link before you act on it or share it. A transcript is what was said; the video is what happened.

What the UserTesting MCP does not return

These are current limits of the UserTesting MCP, not of UserTesting.

Note: Task-level responses and quantitative data are available today through the UserTesting API - see Results API (New experience).

FAQ

Do I need WRITE permission to read transcripts?

No. READ permission on that test type in that workspace is enough. The UserTesting MCP returns results only from tests you can already open in UserTesting.

Can I get transcripts from Classic tests?

No. The UserTesting MCP currently reads results from new experience tests only.

Does the UserTesting MCP return participant demographics or screener answers?

This is not available yet.

Can I get sentiment?

Yes, if you ask for it. Sentiment is off by default.

Can I get the test plan or the quantitative results?

Not yet through the UserTesting MCP. Task-level responses and quantitative data are available today through the UserTesting API.

A quote looks wrong. How do I check it?

Open the video link your AI client returned with the quote. It opens that session's video in UserTesting.

Who do I contact about a problem?

Email support@usertesting.com.

Related content