The accessibility gaps I found in 26 competitors became the rules

Role: Product Designer
Timeline: Feb, Nov 2023
Team: 8 people
Scope: Product design and marketing

The nibbit home page in light theme: a white room with faint perspective lines, the headline 'Take on life a nibb at a time', a purple Create account button, an illustration of a woman sitting cross-legged with a tablet, and a speech bubble quoting the team about planning habits and goals. — home, light, the page the work settled on

Problem

People rarely fail at habits because they picked the wrong one. They fail around week two, when the novelty goes and nothing replaces it. Interviews and a survey kept returning the same two complaints: motivation runs out, and quitting a bad habit gets no help, because nearly every app is built to add. Habbit answered both, building and quitting feeding one system of streaks, levels and trophies, with an AI companion called Pebble to talk to about the progress. The launch site had to carry that promise before any of it existed, and its first concept did not.

firstthe opening concept, the pun lands, the promise never arrives

Research

Habbit was three products wearing one name, a habit tracker, a game, and a conversation with an AI, so the audit covered all three markets rather than the one on the label. Interviews, a survey, personas and journey maps ran alongside it, and usability testing followed on the screens that came out. The habit trackers were the least surprising and the most useful: not one of them was built around the week-two problem.

the audit, 26 competitors across three markets, 16 dimensions each
Market	Audited	What it was there to answer
Habit trackers	6	The direct comparison, pricing, streaks, and where each one loses people
Gaming	7	How returning is made to feel like a reward rather than a duty
AI chat	13	How a companion talks without pretending to be a person

Accessibility

Accessibility was one of the sixteen columns, so all twenty-six products were read for it, and the same failures kept coming back. Those complaints became the rules, for the app and the site alike. The site was then tabbed end to end: keyboard only, no screen reader, and I will not claim otherwise. The palette was measured rather than assumed, the mark carries 16.3:1 on its outline, so the colour is decoration and the black line is doing the work.

390 lightthe narrow width, one control for the nav, reading order unchanged

The same 390-pixel-wide home page in dark theme: near-black background and white text, with the illustration and the purple button keeping their own colour rather than being dimmed with the rest of the page. — **390 dark**the same stack, second theme, the illustration keeps its values

what the audit found in competitors, and what it changed here
Found in a competitor	Rule it became
Titles built as text tags, not headings, in both leading habit trackers	Structure written as structure, so assistive tech gets a real outline
No features for audio or visual impairment, and English only	Short plain copy, and a layout built to take a second language
A text field under half the standard size	Controls sized to be hit rather than aimed at
Dark theme only, noted as making some people uncomfortable	Two themes, drawn separately rather than one inverted

System

The design system reached v2.1 before the MVP shipped, and its cover states the goal in the team's own words: an accessible system with meaningful elements. What makes it usable is that it is drawn for the states nobody enjoys drawing, checkboxes in five, radios in four, every indicator in both light and dark. That last one is why two themes were a build decision rather than a retrofit. Type is split by job, after six candidates were set in one specimen line and compared: Poppins carries the headings, Roboto everything you actually read.

the type scale, two families, split by the job each does
Style	Face	Size / line height
H1	Poppins Bold	60 / 72
H3	Poppins Bold	48 / 60
H5	Poppins Bold	24 / 36
Body large	Roboto Regular	16 / 24
Body	Roboto Regular	14 / 20
Caption	Roboto Regular	12 / 18

Product

Two screens carry the argument. Pebble opens by admitting what it does not know, then offers habits as cards you add with one tap. Each card already carries its category, time and days, so agreeing to a habit and scheduling it are the same action. The success screen is where the gamification and the accessibility meet: completed days are marked with a tick as well as a filled ring, so the state is never carried by colour alone.

The Habbit conversation screen on a phone: Pebble says that since the user did not mention what kind of habit they prefer, it can recommend some life-changing ones. Below sit suggestion cards, Meditate under Mental health, five times a week, at 9pm on Monday to Friday, each with an Add button, and a message field with the keyboard open. — **pebble**the companion suggests, and adding is one tap, habit and schedule together

The Habbit success screen: a chick sitting in a gold trophy with confetti, the heading "Succeed!", and the line "You have completed your daily habit, good job." Under it the habit "Read Homo Sapiens" with a Tuesday-to-Saturday strip where completed days carry both a filled ring and a tick, and a Done button. — **reward**completed days carry a tick as well as a ring, never colour alone

Outcome

Both halves shipped as designs: the research, the system, the app screens, and the launch site and brand around them. Not all of it was finished, the commitments section went into the final About page still carrying a bullet that reads "Community:" with nothing after the colon. And the site itself is gone; nibbit.ai today serves a seven-line poem. The work stands on its judgement rather than its traffic, which is the honest claim for a product that never reached the people it was researched for.

emailthe welcome email, three promises, each with the screen that would keep it

competitors audited, three markets

10,30

people interviewed and surveyed

states drawn per checkbox