How we score websites

StayOrBounce measures the first impression of a website by asking real people four fixed questions about a screenshot of its homepage. Each answer is a score from 1 to 5. Many short reviews together give a reliable average, and the share of reviewers who would stay becomes the stay score. This page explains each step.

What reviewers see

Reviewers see screenshots of your homepage as it loads for a new visitor: one on a laptop screen of 1440 × 900 pixels and one on a phone screen of 390 × 844 pixels. They only see what fits on the first screen, above the fold. Screenshots instead of the live site mean every reviewer judges exactly the same thing, and no one gets stuck on a slow page or a pop-up that only some visitors see.

The four questions

Every review answers the same four questions on a scale from 1 (no) to 5 (yes):

1. Do I understand what this is? Is it clear at a glance what is offered?
2. Do I trust it? Would someone leave their details or money here?
3. Does it look professional? Does it feel polished, or half finished?
4. Would I stay? Would this visitor click on, or leave?

Fixed questions make votes comparable: between reviewers, between your campaigns, and before and after a redesign.

Looking first, then judging

The questions only appear after the screenshot has been visible for at least five seconds, and our server checks that time as well. Each person reviews a website only once, so every vote is a genuine first impression. Reviewers can do up to 10 reviews a day, which keeps the votes spread over many people.

Reviewers only get websites in languages they read. Someone who can't read the site can't judge whether they understand it.

The stay score

The stay score is the share of reviewers who answered 4 or 5 to Would I stay?. It is the headline number of every result. Next to it, you see the average for each of the four questions, so you can tell why people would leave: unclear, untrustworthy or unprofessional.

A result is marked reliable from 10 votes. Below that, you see it as an interim result.

Filtering out unreliable votes

Quick reviews only work if people answer honestly. Two safeguards keep the results clean:

• Automatic-looking votes don't count. A single 5-5-5-5 can be honest, so we only look at habits: someone who gives every question the same score in most of their recent reviews, or exactly the same answers to every website. Those votes are filtered out of the result and earn the reviewer no credits.
• Reliable reviewers count a little more. Once a website has enough votes, we compare each reviewer's answers with the others. Reviewers who usually agree with the group get a weight of up to 1; reviewers who often deviate count for at least half. Honest minority opinions still count, but someone clicking at random barely moves the score.

Target audiences

Website owners can ask for votes from a specific audience: by country, age group or role, such as entrepreneurs or designers. Reviewers fill in this profile themselves. Only reviewers who match get to see the website.

What we don't measure

StayOrBounce measures the first impression of your homepage. It doesn't replace analytics, which tells you what visitors actually do, or a usability test, which shows whether people can complete a task. It answers the question that comes before both: does the first screen convince people to stay?