cover.png

Very first project at Redfin

Home Buying Guide

 

This is the very first project I worked on at Redfin. We conducted two user tests and two A/B tests to get to the final version, which we see a 8.45% lift on the click rate.

When

August 2019, 2020

Role

Duration

Product Designer

3 weeks + 3 weeks

Skills involved

• Redesign driven by data

• Improve content strategy and hierarchy 

The Home Buying Guide page doesn't work hard for our users & our SEO

The problem

As a real estate website, improving the resources page will not only help our users ramp up on the home buying and selling process, but also help up rank higher on Google Search, which leads to more traffic.

Currently, this page has several problems including content hierarchy, readability, and content selection.

For every project, we start with users problem, business goal and metrics.

User problems

  • I’m not sure where to start, and I want general guidance on how to buy or sell a home.

  • I have a specific question about real estate, and I want to find the answer easily—whether that’s from within Redfin or from Google.

Business goals

  • Increase organic traffic

  • Increase unattributed links

Metrics

  • Click rate (short term)

  • Unattributed links (long term)

  • SEO ranking (long term)

old Home Buying Guide page copy.jpg

Research

Knowing users' need as the first step

We conducted 2 rounds of user tests with 20 home buyers and sellers to answer two questions:

  • What content would be most helpful for home buyers and sellers?

  • How do they usually get real estate information today, and why?

Research.png

From the research, we decided the two design principles are easy to read and trustworthy. As for content, we will prioritize the content users need most.

First round

Improving the content and the visual 

Based on the research and several rounds of design critiques, this was the design we shipped.

 

We set a clear hierarchy on this page:

  • A tab to select buying guide or selling guide at the top

  • Featured articles based on users' top need

  • Categorized articles

  • All articles

The result

Soon after, we found out that the click rate went down a lot. It was contrary to everyone's expectation.

We made this page more readable, but less users want to read more

Problem shooting

To figure out what was the problem, we tried different approaches, including user test, live user interview, A/B test with different content. The heatmap was the key to this question. 

From the heatmap, we can see users' attention is not on the content at the first glance. They see the title of the page first, the illustration next. Then it lands on the featured articles.

 

Also, most of the content is not exposed on the first screen. The not-so-important information occupied the most important position.

Heatmap.png

Additionally, I gathered the design pattern from the most-viewed information-heavy websites, from which we can get a sense of the most common patterns - the first screen is the most informative. Users can easily scan and choose the information they want to read.

image 4.png
image 3.png
Always back up your point with research

Second round

With the second round of design, we did two changes

  • Shortened the header
  • Exposed more articles on the first screen

However, my Product Manager and I had different opinions on the two options below.

He prefers the photo option because that's most of our competitors do.

I prefer the illustration option because too many photos make the page look busy and users would not know where to focus.

I ran a user test with 10 users to validate our points.

From the test, we learned:

  • Users can easily find what they want to read on both variants.
  • The illustrations make the page look more formal and professional.

Option 1 - photos

Option 2 - illustration

Although the user test result favored the illustration option, my PM still feel extremely strongly about the photo option. We decided to ship both variants and learn from the real-world data.

Insist on your judgement when you know you are right

Result

2 months later, we got the data back. The illustration variant won, just as the user test showed, and we rolled out it to 100%.

We found that the Illustrations Variant drove an 8.45% lift in click rate over Control and a 6.21% lift over the Images Variant. We'll roll out these changes to 100%.

Version 1

Version 2

The final version!

quote.png

Fan, I especially wanted to send kudos for your insistence that we test illustrations. You were right!! I'm glad you kept standing up for and advocating for illustrations based on what you'd seen in research, and the data has validated that. 

quote copy.png