2022-2023・EdTech・Web・Launched

Chegg / Increased Posted Questions

Led the design for 6 product-led-growth experiments, resulting in a total 1.6M increase in posted questions within a quarter.

Team

1 Content designer, 1 PM, 1 UXR, 2 SWEs, 1 DS

Role

Product design, Product strategy

Timeline

2022.12-2023.2

Context

Chegg’s feature [Posting a question] is the key driver for its business flywheel.

Chegg, a homework help platform with 7M+ college subscribers, relies on the [Post a Question] feature, where users get expert answers.

This feature drives Chegg’s flywheel—more questions → more answers → better SEO → new users → more subscribers. Also, research shows question-posting users are high-value, staying engaged and renewing at higher rates.


Context

Chegg’s feature [Posting a question] is the key driver for its business flywheel.

Chegg, a homework help platform with 7M+ college subscribers, relies on the [Post a Question] feature, where users get expert answers.

This feature drives Chegg’s flywheel—more questions → more answers → better SEO → new users → more subscribers. Also, research shows question-posting users are high-value, staying engaged and renewing at higher rates.


Challenge

Declining question volume and feature engagement

Compared to the previous year, both the number and proportion of questions posted by subscribers have declined.


Challenge

Declining question volume and feature engagement

Compared to the previous year, both the number and proportion of questions posted by subscribers have declined.


Problem statement

How might we encourage subscribers to post more questions on Chegg?

Final Solution

Optimized the E2E posting question flow

  • Made the "Post a Question" (PaQ) entry point more prominent and strategically introduced additional entry points across the platform based on contextual relevance.

  • Streamlined the posting image question flow and improved the usability.



Impact

Successfully driving business growth

Through design experiments, we successfully achieved our key metrics—boosting the number of posted questions and increasing overall feature engagement.


Impact

Successfully driving business growth

Through design experiments, we successfully achieved our key metrics—boosting the number of posted questions and increasing overall feature engagement.


Understand

Identifying barriers to posting

To understand why students weren’t posting questions, I analyzed data to pinpoint drop-off points and conducted additional research




By reviewing analytics on the Post a Question (PAQ) flow, I discovered major friction points:

Understand

Identifying barriers to posting

To understand why students weren’t posting questions, I analyzed data to pinpoint drop-off points and conducted additional research




By reviewing analytics on the Post a Question (PAQ) flow, I discovered major friction points:

Explore

01 Low Awareness

Finding

Most students landed on a Q&A page but overlooked the existing Post a Question widget.

Design variations

  1. Floating button – More noticeable but could disrupt the reading experience.

  2. Banner notification – Highly visible but risked banner blindness.

  3. Widget (Final Choice) – Save one step and can accommodate additional info, but need to make it more visually prominent.

Fine-tuned the visual & content

  • Refined the text hierarchy to make the purpose clearer.

  • Updated the background color and added an illustration for better visibility.

  • Enlarged the input box to encourage students to start typing their questions.

02 Limited Access

Finding

Home page and search page are two other high-traffic area for subscribers, but the old navigation made it hard to find the PAQ page.

02 Limited Access

Finding

Home page and search page are two other high-traffic area for subscribers, but the old navigation made it hard to find the PAQ page.

Search page: Adjusted the widget based on students’ intents

This ensured that students who were less satisfied with the search results were gently encouraged to engage with the Q&A feature. Also, I worked with the content designer to update the contextual copy to acknowledge the students' search experience and provide a more personalized interaction.

  • Exact match: Kept the widget low on the page (to avoid disrupting users who have found what they needed).

  • Great match: Placed the widget below the first answer, encouraging students to ask.

  • Partial match: Placed the widget at the top with a CTA: "Be the first to ask!"

Homepage: Ran A/B/C test to find the ideal widget placement that wouldn’t hurt other features

03 Posting Friction

Finding

From the amplitude data, only 31% of users who reached the PAQ page completed posting. This was a big drop-off.

To better discover those issues, I invited stakeholders to do a dogfooding session, gave them user account and example question to really try to put ourselves in students' shoes.



And from there, we found that:

03 Posting Friction

Finding

From the amplitude data, only 31% of users who reached the PAQ page completed posting. This was a big drop-off.

To better discover those issues, I invited stakeholders to do a dogfooding session, gave them user account and example question to really try to put ourselves in students' shoes.



And from there, we found that:

Proposed Solutions

Deal with eng pushback

I got the pushed back from eng since we could not remove the subject selection step since we needed use that info to assign the questions to expert.

With this constraint, I tried to explore some solutions that still keep the subject selection step but overall streamline the process.

The first idea is to combine the two steps together, however, the subject list would still be long.



After discussing with eng team, I iterated the design by enabling auto-scroll and provide suggested subject list.

Reflect

Thinking beyond just more questions

After I shared the solution with the PM from the expert, I got some feedback:even though we were able to make students post more questions, a lot of those questions were not “really successfully posted” since the image were dark and blurry or it contains more than 1 question in the screenshot so it was hard for the expert side to solve it. 



I realized that we should consider the whole picture instead of just one side. So I suggested some fast follow up items like provide instructions of good image examples and supported file types. Also provide a more direct review and cropping capabilities. 

Learnings

  • Big problems don’t always require big changes. Sometimes, small, thoughtful adjustments can have a significant impact.

  • Dive deeper into both qualitative and quantitive data and find the opportunities.