UX audit

notion.com

81 · Strong

About

Regions

1 / 11

Clickthrough

https://www.notion.com/

Rendering the page. Heavy sites can take up to a minute…

Measuring the capture…

100%

Lens guide

What each view of the capture measures

Every lens below is measured off the capture itself or off the findings. None of them predict clicks or conversion, because that needs real visitors.

Attention

A saliency estimate measured off the pixels: local contrast, colour intensity and edge density. These are the low-level features that decide what the eye is pulled to in the first second, before anyone has read a word.

How to read it: Red is loud, blue is quiet. Check that your primary call to action and your value proposition sit in the red. If a decorative image is the hottest thing on the page, it is winning a fight you did not mean to start.
When to use it: Comparing two designs, sanity-checking a hero, or working out why a button nobody clicks is being missed.
When not to trust it: This is first-glance attention, not clicks and not conversion. It knows nothing about intent, copy or your audience, a boring but expected button can outperform a loud one.

Legibility

Cells that contain fine detail, the signature of text, but where the light-to-dark spread is too small for that detail to separate from its background.

How to read it: Every red patch is somewhere a reader will squint. Zoom into one and you will usually find grey-on-grey body copy, a caption over a photo, or placeholder text in a form field.
When to use it: Accessibility passes, dark-mode reviews, and any page where the copy matters more than the picture.
When not to trust it: It measures the rendered image, so it cannot quote you an exact WCAG ratio for a specific text colour, and it will flag deliberately faint decorative textures alongside real problems.

Density

Visual density: how many distinct elements, edges and boundaries are packed into each part of the page.

How to read it: Hot areas are crowded, cold areas are calm. A good page has rhythm: dense blocks separated by quiet ones. A wall of uniform heat means nothing is being prioritised and the reader has to do the sorting.
When to use it: Diagnosing pages that feel overwhelming, and finding sections where whitespace would do more than another feature card.
When not to trust it: Dense is not automatically bad. Pricing tables, dashboards and comparison grids are meant to be dense; judge them against their purpose, not against the colour.

Fold

Where the bottom of the screen falls for the desktop sizes most visitors actually use, drawn straight onto the capture.

How to read it: Everything above the first line is what a visitor sees before they do anything at all. Ask whether the promise and the next step are both above that line on the smallest of the three.
When to use it: Hero reviews, landing pages, and any argument about whether something needs to move up.
When not to trust it: These are desktop lines only. The capture is rendered at desktop width, so drawing a phone fold on it would be fiction. Check mobile on a real device.

Flow heat

The regions that carry the reconstructed flow selected on the left, weighted by how much of that journey depends on each one.

How to read it: The hottest region is the one the flow leans on hardest, and usually the one worth testing first. Cold or missing regions play no part in this particular journey.
When to use it: Working through one journey at a time (signup, checkout, contact) instead of the page as a whole.
When not to trust it: The flow is reconstructed from the markup, not recorded from real sessions. It is a plausible path, not a measured one.

Links

Every anchor found in the markup gets a numbered, colour-coded dot inside the region it belongs to, and the readable list of what each number is sits beside the capture, outside the frame. Colour is the kind of exit: call to action, navigation, in content, footer, off site, same page, email, phone. Clicking a dot or the arrow on its row runs a full audit of that destination and adds it to this session as the next page. Tick the boxes instead and prowl them: each one gets a capture on the session map, no audit spent, so you can see where the site goes before deciding what is worth a scan.

How to read it: Read it as the page’s exits. A region with no links is a dead end; a region with twenty is asking the visitor to choose instead of telling them what to do next.
When to use it: Walking a journey (landing to pricing to signup) and auditing each step in the same workspace.
When not to trust it: A dot marks the region an anchor belongs to and its place in reading order, not its pixels: the scan reads markup, so placement inside the band is an estimate and should be read as “somewhere in here”, never as a measured coordinate. Same-page jumps, email and phone links are annotated but cannot be audited, and they are marked as such. Anchors added by JavaScript after load are invisible to a static read.

Vision simulation

These filters show the capture as it would appear to someone with a visual impairment. Colour vision lenses use the Machado, Oliveira & Fernandes 2009 model. Field-loss lenses are labelled as illustrations.

Deuteranopia

Red-green colour blindness (green cones absent)

Affects roughly 1 in 12 men and 1 in 200 women.

Protanopia

Red-green colour blindness (red cones absent)

Affects roughly 1 in 100 men.

Tritanopia

Blue-yellow colour blindness (blue cones absent)

Rare, roughly 1 in 10 000 people.

Achromatopsia

Total colour blindness, only luminance remains

Rare, less than 0.001% of the population.

Low vision

Uncorrected refractive blur (approx. -2 dioptres)

Roughly 2.2 billion people globally have some form of near or distance vision impairment.

Aging eyes

Yellowed lens, reduced contrast and extra glare past age 65

A normal part of ageing affecting everyone over 60 to varying degrees.

Cataracts

Clouded lens with severe contrast loss and scatter

Affects roughly half of people over 75.

Tunnel visionillustration

Peripheral field loss, as in advanced glaucoma

Glaucoma affects roughly 3% of people over 40.

Macular deg.illustration

Central field loss, as in age-related macular degeneration

Affects roughly 8% of people over 45.

Verdict

/ 100 · Strong

This page broadly does the job of a landing page by making the primary conversion visible immediately with “Get Notion free” and a secondary path to “Request a demo.” The main weakness is message clarity: the hero stacks several concepts at once — “Where teams and agents Think together,” “Capture context, find answers, and automate tasks,” plus multiple AI product labels in the nav — so a first-time visitor has to decode what Notion is before deciding whether to act. For a page whose job is one conversion, that ambiguity costs intent at the top of the funnel.

Scored on the flat rubric

This report was produced before we started reading what a site is for, so every finding counted the same. Re-run the scan to score it against its purpose.

What to say first

Judged as a Landing page(likely)

site root · A marketing page: its job is one conversion.

8 observations · 6 regions · 3 flows