Building An Online Tool That Helps 150M+ Second Language English Speakers In Writing English with Ease

4 min readNov 24, 2023

QuillBot is an advanced AI-powered online writing tool. It makes use of cutting-edge natural language processing (NLP) techniques to help users write better, come up with new phrases, and improve the clarity and quality of their content.

AI / SAAS sector | 2 months timeline | Web app designing | B2B & B2C

My role as a UI/UX designer involved cross-team collaborations, research & design.

Before I came in,

QuillBot existed with around 1M users, a WIP web app with just 2 features paraphraser & summariser. The payment flow, the dashboard’s peripheral area & payment flow were all under the ideation stage.

The problem is,

Non-native speakers of any language expect guidance. English, with wide usage, is a need in this era. Many non-native speakers struggle to write college essays & work emails. Also, they struggle to learn the language. So the process of writing is time-consuming & difficult for them.

Competitors were focused on just correcting the text, instead of educating the users, which also became our opportunity.

We followed simple 3 step analysis process

Attitudinal analysis

Interviews, Survey, Online Feedback
92% of users expressed satisfaction with QuillBot’s paraphrasing accuracy. However, 15% suggested the need for more clearer UI of tools for undisturbed writing. 78% of respondents found the summarization feature helpful for condensing lengthy texts, while 10% requested more options for customizing summary length.

Behavioral analysis

Heatmap, Feature adoption, Bounce rate study
73.3% was the average adoption rate for the 3 features. We also found that the bounce rate was low (18%), but the reasons were strong.

1. Confusing User Interface: 40% of bounce instances occurred due to users finding the UI difficult to navigate, leading to frustration and abandonment.

2. Slow Response Time: 30% of bounces were attributed to slow response times, especially during peak usage hours, causing users to leave in search of faster alternatives.

3. Inaccurate Results: 20% of users bounced because they found inaccuracies in the paraphrased or summarized content, suggesting issues with the algorithms.

4. Lack of Customization: 10% of users left due to a perceived lack of customization options, indicating a need for more tailored user experiences.

Quantitative analysis

Data analysis via Amplitude

Session Time:
Average Session Duration: 9 minutes
Insight: Users spend considerable time on QuillBot, indicating sustained engagement with the platform’s features.

Drop Rate:
Overall Drop Rate: 12%
Insight: A moderate drop rate suggests most users stay engaged, but improvements could reduce the abandonment rate further.

Average Daily Users:
Peak Usage Hours: 10 AM — 12 PM and 7 PM — 9 PM
Average Daily Users: 1.2 million
Insight: QuillBot experiences high user activity during these peak hours, indicating its popularity during work and study times.

Feature Usage:
Paraphrasing Tool Usage: 65% of total users
Summarization Feature Usage: 45% of total users
Grammar Checker Usage: 55% of total users
Insight: The paraphrasing tool is the most popular, followed closely by the grammar checker, while the summarization feature lags in adoption.

Time Spent per Feature:
Paraphrasing Tool: 10 minutes on average per session
Summarization Feature: 6 minutes on average per session
Grammar Checker: 8 minutes on average per session
Insight: Users spend varying amounts of time on different features, with the paraphrasing tool having the highest user engagement.

User Retention Rate:
30-Day Retention Rate: 40%
Insight: 40% of users continue to use QuillBot after 30 days, indicating a decent level of user satisfaction and long-term engagement.