About Feedbaq

Why it exists

Analytics can show that someone abandoned onboarding or stopped using a feature. Talking to them helps explain why. But arranging interviews takes time, and a short survey answer can leave you with more questions than answers. Feedbaq makes it easier to collect detailed feedback without scheduling a call with every participant. You share a study link, people respond when it suits them, and you get their explanations alongside the recordings and transcripts.

Research the decisions you need to make

Use Feedbaq to explore a problem before building, test a product or design concept, understand an onboarding experience, or find out why customers cancel. Designers can ask people to walk through a prototype and explain what they expect to happen. Marketers can put landing page copy or a message in front of people and ask what they understand, what feels relevant and what would stop them from taking the next step.

Start from a template, write your own questions or ask your AI agent to draft a study around the decision you are making. Review the questions, publish the study and share its link with the people you want to hear from. Participants answer in their browser without creating an account.

Hear the explanation and see the experience

A spoken answer gives someone room to explain the situation in their own words. A narrated screen recording lets you watch them use a product, navigate a page or try a task while they tell you what they are thinking. Screen recording works in a desktop browser and captures both the screen and their voice.

You can combine voice, screen recording, text and multiple-choice questions in the same study. That means you can collect a clear choice and then ask for the story behind it, or ask someone to show you a problem instead of trying to describe it from memory. You choose the answer formats for each question.

Ask about what each person actually said

With AI follow-ups enabled, Feedbaq can ask a new question based on the answer it just received. If someone says onboarding was confusing, a follow-up can ask which step was unclear. If they describe a workaround, it can ask how they do it today.

You write the main questions, choose where follow-ups belong and can give the AI instructions for each question. The next question can respond to that person's experience, rather than being limited to the questions you anticipated when you built the study.

Run the study with your own AI agent

Feedbaq connects to Claude, ChatGPT and other compatible AI agents through MCP. Your agent can create and edit a study, configure its questions and follow-ups, publish it and read the full responses. If it already has your product context, research goals or earlier findings, you can use that context to shape the next study.

For example, ask your agent to draft a study for customers who cancelled, review it in the builder, then share the published link. When answers arrive, ask the same agent to compare the reasons people gave and pull out quotes that explain the patterns. The MCP guide explains how to connect it.

Go from answers to evidence you can check

Spoken answers and screen recordings come with transcripts. Word timings let you jump from the transcript to the moment in the recording, so you can hear the tone or watch what was happening. Feedbaq also labels open answers by topic and shows topic counts and highlighted answers in the results, helping you find recurring issues and useful examples.

Your connected AI agent can read the full responses, compare what people said and quote answers with links back to the evidence. You can follow a finding back to the participant's own words before deciding what to change.

Who makes it

Peter Kildegaard founded Feedbaq. Before Feedbaq, Peter worked as a staff product manager at Maze, the product research platform, helping customers such as Anthropic, Revolut and Ramp understand their customers. Read Peter's guides.