ai.google website redesign, discovery, usability • 2024

Redesigning the ai.google website for tech-forward AI consumers and influencers

Overview: Ahead of Google I/O in 2024, my team worked with Google AI to redesign their website. Working closely with a strategist, producer, and two designers, I led two rounds of research - the first focused on discovery and audience definition, and the second on usability and function.

My responsibilities: I led all research efforts including research design, recruiting a niche audience, synthesizing results into actionable insights, and sharing them directly with our key stakeholder at Google.

My team: 1 Researcher (me), 1 Strategist, 2 Designers, 1 Producer

The text-heavy, repository-style site failed to tell a cohesive story, showcase proof points, and illustrate Google’s commitment to responsibility.

KEY PROBLEM

Despite Google’s history and deep experience and expertise in AI the existing ai.google site failed to excite users about products, and instill trust and confidence in their approach.

It was thorough, highlighted dense principles & objectives and dove into the details of AI responsibility rather than allowing consumers to understand relevant applications.

Through conversations with stakeholders and other teams at Google, we determined that we needed to design a site that would differentiate Google from its competitors and:

Tell Google’s story as a legacy AI leader

Screenshot of a webpage titled 'Our Principles' under the section 'Objectives for AI applications,' with a description and several bullet points outlining AI principles.

Excite consumers about AI products

Build trust around AI safety & responsibility

In support of these goals, I led two rounds of research: discovery and usability

PROCESS

Through primary user research with our core audiences, I needed to:

  • Understand audience perceptions of the AI landscape, including current and historical leaders in the industry

  • Evaluate attitudes around AI responsibility and safety

  • Evaluate how participants respond to existing Google and competitor websites (Microsoft, OpenAI, Meta)

  • Once designed, test usability of the updated website

Laying strategic foundations

I conducted 10, 60 minute in-depth interviews via Lookback.io with tech-forward consumers and influencers. Sessions covered AI perceptions, reactions to 3 competitor website (Microsoft, OpenAI, MetaAI), and impressions of ai.google.

Recruitment Note: We initially targeted tech influencers who publish content to auxdiences of 10K+, but found them very challenging to recruit in a tight timeline. Rather, we spoke with proxy influencers, and scoped more time to recruit “true” influencers for phase 2.

Screenshots of three websites about AI and technology, including Microsoft AI, OpenAI, and Meta, showcasing AI research, tools, and communication.

PHASE 1: DISCOVERY

Insights & Recommendations

PHASE 1 DELIVERABLES

Digital screenshots of organized and color-coded project management or brainstorming boards with various categories and notes.
Slide presentation with slides discussing opportunities and key findings related to AI, Google, and humanity. Topics include expanding perceptions beyond generative AI, leveraging Google's scientific legacy, trusting AI, and emphasizing specific approaches and examples.

Following data collection, I ran a collaborative team workshop to begin to co-create themes with teammates who observed sessions. Together, we began to ideate on key insights and tie them to design recommendations. 

WORKSHOPPING & SYNTHESIS

REPORTING & PRESENTATION

After the workshop, I dove deeper into synthesis and ultimately created and presented a report that outlined audience themes, key needs and pain points around audience needs for the website, and design recommendations. 

DESIGN RECOMMENDATIONS

Though the report went into more depth, the key opportunities for design can be summarized into 5 simple recommendations:

  • Work to expand understanding of AI beyond LLM applications

  • Embrace specificity

  • Fold in scientific legacy in highly relatable ways

  • Position Google as the responsible, safe leader

  • Show (don’t tell!) the relationship between AI & humanity

The design team ran with these recommendations and strategic foundations to redesign the information architecture and key pages, ahead of the second phase of testing.

ADDITIONAL FINDINGS OF NOTE (click to expand)

    • AI = generative AI. For our audience, AI is synonymous with generative AI despite pre-LLM technology. (Rare) highly technical influencers are outliers.

    • OpenAI is the leader. They dominate public mindshare; participants see Google as cautious & late to the game despite historical contributions.

    • Hopeful but skeptical. Though users see AI as helpful, they need proof of responsibility and see the landscape as lacking in this area.

    • Specificity. Participants emphasize desire to see & interact with tools and products to garner excitement.

    • Relatability. Participants look for examples that relate to their lives and desired use cases.

    • Innovation and responsibility. Across the industry, companies have room to better communicate how they’re pursuing safety and responsibility.

PHASE 2: USABILITY

Validating Designs and Assumptions

Collection of web pages and UI design mockups related to Google's Gemini AI ecosystem and AI platform, showcasing text, images, and interface elements.

After a few weeks of design iteration rolling in the strategic principles from discovery research, I returned to the project and conducted 10 additional 60 minute in-depth interviews.

These sessions focused on reviewing clickable Figma prototypes of the redesigned ai.google site. Pages included:

  • Redesigned homepage

  • “Meet the Models”

  • Gemini Ecosystem

  • Navigation

In interviews, I focused on comprehension and resonance of the content across the pages including visual design, written content, and storytelling.

AUDIENCE ADJUSTMENTS

In Phase 2, we validated audience findings from “proxy influencers” with “true influencers"

As we had trouble recruiting influencers with audiences of above 10K followers in phase 1, I spent the break between research phases creating a recruitment toolkit, hand-sourcing, and using email to personally reach out and recruit members of the audience.

Ultimately, we found that website needs were consistent across what we’d uncovered with the proxy group, though uncovered additional learnings about how influencers learn and talk about new AI tools as they come to market.

Two overlapping pages with a document titled 'Google AI - ai.google re-design Influencer Recruitment Kit,' containing a table of the recruitment timeline and sections for participants and additional information.

PHASE 2 DELIVERABLES

Tactical Design Recommendations

After an additional round of workshopping and synthesis, I created a report that outlined findings mapped to specific design recommendations for each of the pages and features that we tested.

Flowchart comparing perspectives on AI messaging, integration, and communication strategies for developers and consumers.

PROJECT IMPACTS

The 2 research sprints informed the project at both the strategic and tactical UX levels

Taken together, the two sprints informed the design of ai.google, particularly by

  • Understanding the key audiences needs, mental models, and

  • Simplifying and humanizing overly technical parts of the website

  • Shifting tone from corporate to human

  • Elevating key responsibility & safety initiatives in more engaging ways

Beyond the specific and tactical improvements, this work built client confidence in our team. They deeply appreciated our persistence in reaching a specific, niche influencer audience in Phase 2, and were impressed by how well our proxy group represented the attitudes and perceptions held by the true target users.

Screenshots of technology-focused webpages about building beneficial AI, Google AI updates, and advancements in AI research.

REFLECTIONS

A key challenge in this research was recruiting influencers with large public audiences in Phase 1. Pivoting to speak with “proxy influencers” - people with smaller but meaningful audiences, either within companies or on LinkedIn - allowed us to gather early insights that quickly informed design. While the team iterated on prototypes, I was able to recruit higher-impact influencers to validate those initial findings, ensuring our insights held true for the broader audience.

This experience reinforced the importance of flexibility in recruitment and the value of validating early insights with the most representative participants whenever possible.