How to get honest feedback on your website

The people easiest to ask, friends, colleagues and other founders, know too much and want to be kind. Open questions get opinions about colours. And feedback from people who are not your customers is advice about a different business. Good feedback needs the right reader, a concrete task, and some way to tell a pattern from one person's taste.

• Six methods compared: who gives the feedback, what it needs, what it misses.
• How to ask so people tell you what is wrong, not what is nice.
• Which methods need traffic, and which work before launch.
• How to tell a pattern from one person's preference.

Free and fast, and good for typos, broken links and obvious confusion. The problem is that they know what you do, and they are on your side. They will rarely tell you that the headline means nothing to them.

Get more out of them by giving a task instead of asking for an opinion: "Find out how much this costs", or "Tell me in one sentence what this company does". Then stay quiet and watch.

Founder forums, subreddits, and Slack or Discord groups for your niche can give blunt, outside feedback, and some have threads set aside for it. Read each community's rules on self-promotion first, and ask a specific question.

The catch is who answers. Community members are often builders and marketers rather than your customers, so their feedback leans toward craft: design polish, copywriting rules. Useful, but not the same as a buyer's reaction.

Tools in this category record what real visitors do on your site: how far they scroll, where they click, where they hesitate. Hotjar, now part of Contentsquare, is a well-known example. They show real behavior at scale, and they are the best way to find where people struggle once you have traffic.

They need traffic, they show what people did rather than why, and someone has to watch the recordings. Pair them with a question asked at the right moment.

Hotjar vs Mimiq

A one-question survey on exit or after an action, such as "What almost stopped you from signing up?", gets real reasons from real visitors. Few people answer, and the ones who do are self-selected, but the answers are often specific enough to act on.

Ask one question at a time, at the moment it matters, and read every answer rather than charting them.

Watching people from your audience try your site, moderated on a call or unmoderated through a research platform, is the strongest evidence of where real people get confused. Platforms such as UserTesting, Maze, Lyssna and Useberry recruit participants or let you bring your own.

It costs money and time to recruit, run and watch. It is the right tool for decisions that are expensive to get wrong, after cheaper methods have cleared the obvious problems.

Mimiq vs Lyssna Mimiq vs UserTesting

Simulated people give outside feedback with no traffic and no recruiting. In a Mimiq page test, 25 simulated people matched to your audience react to your live page in about 90 seconds, and their objections come back ranked by how many people raised them, with each person's reasons. In a flow, they try a task on your site in real browsers and show where they got stuck.

They are built to say no, so you get the objections friends will not give you. They are also simulated: not your customers, weaker for niche experts and lived experience, and not a forecast. Use them to find what to fix and what to ask real people.

Open a sample report

Give a task, not a request for opinions. "What would you do next on this page?" or "Where would you find the price?" produces evidence. "Do you like it?" produces politeness.

Ask for the first impression before anything else, because it disappears once people look around. Then ask what would make them leave, and what would make them trust it.

Look for patterns. One person disliking your colours is taste. Several people missing the price is a finding. Fix what repeats, starting with what sits on the first screen.