Screen recording feedback: see why users get stuck

What screen recording feedback is

Screen recording feedback is feedback people give by sharing their screen while they use something and talking you through it. You get the recording of what they did and a transcript of what they said, synced. Researchers call the talking part thinking aloud. Done without a researcher on the call, it is an unmoderated usability test.

Words tell you what. The screen shows you why.

When someone says they could not do something in your product, you learn what went wrong. When you watch them try, you learn why. “I couldn’t connect my data” could mean a missing feature, a bug, a confusing label, a permissions problem or a tired person on a Friday afternoon. Each needs a different fix. The words cannot tell you which. The screen can.

This is the oldest rule in usability research. Nielsen Norman Group’s version: watch what people actually do, and do not rely on what they say they do. Self-reported data, the same article notes, is several steps removed from what really happened. A recording is not.

The screen gets people talking

People react to what is in front of them, so a screen recording draws out far more than a question asked from memory. Ask someone to describe your onboarding and they give you a summary. Put them in it and every screen prompts a reaction: “Oh, I didn’t see that”, “Is this the same as the other one?”, “This is where I gave up last time.” They are not trying to remember. The product is reminding them.

Contextual research has shown this for decades. Nielsen Norman Group notes that people can explain what they do and why while they are doing it, but when they summarize it later, the reasoning and motivation drop out. Karen Holtzblatt and Hugh Beyer, who created contextual inquiry, found that work becomes so habitual that people are not consciously aware of it until they are in the middle of it.

That is why Jakob Nielsen calls thinking aloud perhaps the single most valuable usability method. You hear what people believe as they act on it, including the wrong beliefs that make them click the wrong thing. It also makes answering easier. Talking over a screen is a more natural, more interactive thing to do than writing a paragraph about it afterwards.

Five people, not five hundred

Jakob Nielsen’s model, from 2000, estimates that five users find about 85% of usability problems, and recommends many small rounds over one big study: fix what five people found, then test again.

Five is a floor, not a ceiling. When Laura Faulkner tested 60 users and drew random groups from them, groups of five found between 55% and 99% of the problems, and every group of ten found at least 80%. With Feedbaq the extra five cost you one more link, not another week of calls.

You do not need to be in the room

Jeff Sauro of MeasuringU compared moderated and unmoderated studies of the same tasks. On average, completion rates and ease ratings came within about a point of each other, and he called the two methods very similar. The difference in effort was not similar at all:

What a moderator adds is the follow-up question. Nielsen Norman Group names this as the main gap in unmoderated testing: there is no chance to ask about what the person just did. Feedbaq’s AI follow-up narrows that gap. When the recording ends, the AI reads what the person said and asks about it, so a comment like “this is where I gave up last time” gets its story.

Dedicated usability testing platforms have been priced for enterprises, on annual contracts that often run to tens of thousands of dollars. In Feedbaq, screen recordings come with the plan at $49 a month, with 600 minutes of recording each month: enough for sixty 10-minute sessions, or twelve rounds of five.

What to put in front of people

Anything on a screen can be tested this way, from a sketch in Figma to a competitor’s checkout.

• Test competitors before you design. Nielsen Norman Group calls competitors’ sites the second-best prototypes of your new design. Someone else already paid to build them. Watch your future customers use them.
• Ask for bugs on video. In a survey of 156 developers, the steps to reproduce were the most wanted part of a bug report, and wrong or missing steps the biggest problem. A recording has the steps.

How screen recording works in Feedbaq

Feedbaq is a tool for collecting feedback from your users. You write a few questions and share them as a link. People answer by voice, screen recording or text, and you get every answer with a transcript. Your AI agent can set it up and analyze the answers for you.

A screen recording is the answer format for anything people do on a screen. The question asks them to share their screen and talk you through a task, for up to 10 minutes:

• Screen and voice, never the camera. Feedbaq records the screen or window people choose, with their narration. It never asks for a camera, so people show you their work, not their face.
• Nothing to install. It works in Chrome, Edge, Safari and Firefox on a computer. For people on a phone, you choose: offer another way to answer, skip the question, or ask them to switch to a computer.
• A task link. Add the page you want tested, whether it is your prototype, your site or a competitor’s, and it opens in a new tab when they start.
• Recording and transcript, synced. Watch the recording with the transcript running alongside, and jump to any moment.
• A follow-up at the end. Turn on AI follow-up questions and the AI asks about what the person said while they worked. They answer by voice or by typing.
• Read by your agent. Your AI agent can read every narration transcript and pull out where people got stuck across all the recordings.

Watch five people this week

1. Pick the one flow you are least sure about. Onboarding, pricing or the feature you shipped last month.
2. Write one task in the words your users would use, not the words on your buttons. The usability walkthrough template gives you one recorded task with narration and short questions about the hardest part.
3. Send the link to five people, on a computer.
4. Watch the recordings. Fix what at least two of them hit, then send it to five more.

Building a study is free. You subscribe when you publish: $49 a month, with 100 responses and every feature included.

Questions people ask

Sources

1. Nielsen Norman Group, “First rule of usability? Don’t listen to users” 2001.
2. Jakob Nielsen, “Thinking aloud: the #1 usability tool” Nielsen Norman Group, 2012.
3. Nielsen Norman Group, “Contextual inquiry” 2020.
4. Karen Holtzblatt and Hugh Beyer, “Contextual design” The Encyclopedia of Human-Computer Interaction.
5. Jakob Nielsen, “Why you only need to test with 5 users” Nielsen Norman Group, 2000.
6. Laura Faulkner, “Beyond the five-user assumption” Behavior Research Methods, 2003.
7. Jeff Sauro, “How do moderated and unmoderated usability tests compare?” MeasuringU, 2018.
8. Jeff Sauro, unmoderated testing time per participant MeasuringU, 2012.
9. Nielsen Norman Group, “Remote usability tests: moderated and unmoderated” 2013.
10. Nielsen Norman Group, “Competitive usability evaluations” 2014.
11. Bettenburg and colleagues, “What makes a good bug report?” FSE 2008.
12. Vendr, usability testing platform contract data Checked 30 September 2026.