The New York Times
Re-invent reading news online
Overview
I am an avid reader and I can tell you I am not a fan of reading News online. Sometimes I open an article and close it with gusto. Sometimes I need an overview of the entire topic instead of just the latest and other times I only want to see what’s latest without the historic context. If I am facing this, I am sure others out there have the same issue.
Print media has revolutionized the world in many ways, online media hasn’t impressed the most yet.
This is a capstone project for which I chose to assume New York Times as a client and decided to add features that improve user experience of reading News.
TL;DR
Problem
While our goal is clear, the approach and solutions are not. There has naturally been a shift towards short videos to consume information as revitalizing reading online information is an ambiguous problem and has a very large user base to account for. The challenge lies in understanding user pain points and identifying prioritized areas of improvement while keeping personal biases out of sight.
The article size does not match the user’s level of interest
Reading news is personal, but News is not personalized
Proposed Solutions
We want to personalize the News based on users’ interests. Making the online news reading experience personalizable requires revitalizing the way we seek information online and solutions might be resource-consuming in a world where News outlets are already under economic pressure.
Fortunately, there is another tool being popularized right now. ChatGPT and Bard. LargeLanguageModels (LLMs) and AI
To get the parity of features with Google Podcast, I decided to also incorporate converting summaries created by AI and LLM into audio
Project Type
Capstone project
Category: Add a feature to an existing successful product
Role
UX Researcher
UX Designer
UI Designer
Visual Designer
Goal
Improve the News reading experience that results in more reading engagement
Timeline
3 weeks
Reading news is personal, but News is not personalized
Reading News is a common activity for people from all across the world, but no two people would have the same interest profile and the current News setup does not account for this fact. This makes the current News setup not a 100% fit for 100% of the readers.
Making news personalize can be resource-consuming in a world where News outlets are already under economic pressure. Thankfully, a new force is on the horizon. Generative AI and LLMs (Large Language Models) to fill that shoes.
Discover
Problem Statement:
While our goal is clear, the approach and solutions are not. There has naturally been a shift towards short videos to consume information as revitalizing reading online information is an ambiguous problem and has a very large user base to account for. The challenge lies in understanding user pain points and identifying prioritized areas of improvement while keeping personal biases out of sight. Clearly, user input was necessary.
Goal
Understand why users read certain articles and skip others
Understand how much time users spend on an article and how it changes over categories of article
Understand user motivations behind reading or not reading News articles
User Interview
I interviewed 5 people for this. 4 males and 1 female, ages ranging from 26-39, all located in North America across USA and Canada, from at least 2 ethnic groups.
Each interview averaged around 45 minutes and 80% of the interviewees shared that reading online News is necessary but tormenting. Most of the time, interviewees cannot find the exact information they care about and 100% of the interviewees confirmed that they open an article and close it without reading it fully
Needs:
Users require overall reports for ongoing events.
They seek specific information from articles.
Users want to cover more news in less time.
They are interested in knowing important information for current topics.
Users desire to consume information that has a personal impact, such as immigration-related news or local events.
Motivation:
Users aim to stay up-to-date with current events.
They want to know important information about news articles quickly.
Users aim to cover more news in less time.
They prefer multitasking while consuming news.
Frustration
Users face time constraints when reading news.
Lengthy articles are time-consuming and lead to information overload.
Users lose interest after reading a few paragraphs and miss important details.
They are unable to find specific information in lengthy articles.
Lengthy articles with noisy and unnecessary data demotivate readers.
Too many paragraphs and advertisements in the news app disrupt the reading flow.
Users can't access overall reports for ongoing news.
Competitive analysis
There are no direct competitors who have tailored their features towards News apps like NYT. Certainly, there are no apps that have integrated variable-length summaries based on users interests. This will give NYT a competitive edge that will easily be converted to more traffic and downloads leading to more business value.
One competition that has a feature that is still unaccounted for in the current agenda was the audio feature from Google Podcast (and other Podcast apps) that allows users to multitask. This was brought up by people who interviewed as well.
Define
Affinity mapping
Next step in the journey to identify the most pressing user pain point, I started organizing and categorizing this data. I used affinity mapping as it allows me to gather similar-looking ideas into the same bucket.
While performing the affinity mapping, it became crystal clear to me that the most pressing issue for users is the size of the article itself. All users find it either too long and time-consuming and others think it’s noisy.
POV Statements
Using the information obtained from the interviews and affinity mapping, I had my mind set on solving users' information fatigue.
I would like to explore ways to help everyday readers provide more summarized articles because summaries will help them absorb more news in less time
I would like to explore ways to help frequent readers provide an overview of the ongoing event because readers want to stay up to date with what’s going on around them.
I would like to explore ways to help busy readers by providing more ways to absorb news because the user wants to multitask when absorbing the news.
HMW Questions
How might we help everyday readers absorb more news in less time?
How might we provide summarized articles to everyday readers?
How might we help frequent readers to stay up to date with current topics?
How might we provide an overall summary of ongoing topics to frequent readers?
How might we help busy readers absorb more news in less time?
How might we provide a structured summary to people who want to multitask during news reading?
Persona
After User Interviews when I got into thinking about ‘who’ would use the feature and ‘why’, one persona came to be prominent.
Develop
User & Task Flow
Summarization is a core feature for LLMs and using LLMs can help reduce and increase the length of the article. It also can help answer questions and provide historical context when trained with the right data set.
For this project, we will assume that NYT has collaborated with some LLMs and will be using LLMs to provide different lengths of summary that users can read based on their interests. So our design goal is to provide a UX that allows users to read different sizes of summaries for a given article based on their interest
For this feature, we will provide a new button on the article page that will take users to a new summary page. On this page, users will be able to read the entire article or variable size summary where users will be able to select size.
Design pattern
The next step was to understand NewYorkTimes app design patterns so that I can create a consistent design with it. I created hand-drawn scratches based on that. Here are some examples
Idea Generation
To summarize my discovery process, I associated my HMWs with the feedback I received from interviews related to the News reading experience. This brought up an easy observation that was hidden all along for a long.
The article size does not match the user’s level of interest
Reading News is a common activity for people from all across the world, but no two people would have the same interest profile or reading style, and the current News setup does not account for this fact. This makes the current News setup not a 100% fit for 100% of the readers.
Reading news is personal, but News is not personalized
We want to personalize the News based on users’ interests. Making the online news reading experience personalizable requires revitalizing the way we seek information online and solutions might be resource-consuming in a world where News outlets are already under economic pressure.
Fortunately, there is another tool being popularized right now. ChatGPT and Bard. LargeLanguageModels (LLMs) and AI
To get the parity of features with Google Podcast, I decided to also incorporate converting summaries created by AI and LLM into audio
Deliver
High fidelity wireframes
With all the UX design work that I have done till now, I was ready to create final UI designs and documents that I could deliver to the development teams. Here is my final UI design that I feel captures both the goals of allowing users to read different-sized summaries and also converting that to audio for listening while multitasking
Measuring success
After such a feature is implemented, it is important to get feedback from the live users. But there are some other ways we can measure the success of this feature.
Measure the number of times users are using the summary feature page for any article
Measure the number of times users are reading small and medium size summary
Usability testing
I created a user testing using Maze and provided it to a few friends and also the 10 people who I interviewed as part of the UX study. Here are some of the highlights from the Maze usability testing.
Average time to complete the task: 13s
100% of the testers provided positive feedback on the design. No iteration needed for now
80% of the testers said that they “strongly prefer” to use such a feature, while others said that they “prefer” to use such a feature if available
100% of the testers said that the design was easy to follow and they were able to complete the tasks easily
Latest Prototype
Next Steps…
Default summary page based on profile
A profile setting can be added that allows users the ability to use the summary page as a default page when a user opens an article. This will provide users with easy access to the summary page if they would like.
Default summary size based on interest
This is a feature that we neglected in the first iteration due to additional work on the development team. Users would have different interests based on their tastes, location, and personal identity. The default summary size should change based on user interest. This would help avoid one more click for users to get to where they want to be. Of course, if the user wants to read different size summary, then they can with that one click.
Q&A System
The most important other user feedback from the interviews was that it is hard to find specific information in the articles. Using AI and LLMs can help with that as well. A new feature to create a Q&A system can be added to the NewYorkTimes app that can answer specific questions from the users.