Research

/

2024–2025

Meta Connect: Building Meta’s Biggest Event Website

I led UX strategy for Meta Connect 2024 from invite to replay, then tested the site with real people to fix what it got wrong.

Project details

Company

Meta

Role

UX strategy and research lead

Methods

UX strategy, information architecture, content strategy, post-launch analytics, attendee survey (n=80), unmoderated usability study on dscout (n=13)

Year

2024–2025

Overview

The Problem: Meta Connect is Meta’s biggest launch event. Its website had to serve invited guests and a global online audience across four phases, built by outside partners and internal teams on two platforms, and nobody owned how it fit together.

The Method: I owned the UX strategy: one site structure and a page-by-page blueprint every team built from. After launch, I tested the public homepage with 13 people on dscout.

The Core Insight: People understood what Connect was, but not how to take part: in person, online, or both.

The Impact: 783K visitors, 225K registrations and 17K product clicks. My framework was reused for Connect 2025 and 2026.

Team: UX strategist on Creative X’s Connect team, working with Jack Morton, RainFocus, Meta AI and Reality Labs.

What made it hard?

Two audiences, four phases, two platforms.

Invited guests needed a private site and an attendee portal. Everyone else needed meta.com/connect. The site changed every phase, from invite to replay, and ran on Meta’s CMS and RainFocus. Every hand-off was a place to lose people.

Diagram of how the synthetic panel was built: 400+ real people applied, 135 were screened, and 100 digital twins were built 1:1 from real interviews.
What did I design?

One blueprint every team built from.

I wrote the page-by-page structure for every phase and both platforms, and changed the main action as the event moved: Register, View Program, Watch Now, Watch On-Demand.

“At times I joked he was able to read minds at how well he understood needs and translated that into a cohesive plan.” Janelle Mungo, Connect

Diagram of how the synthetic panel was built: 400+ real people applied, 135 were screened, and 100 digital twins were built 1:1 from real interviews.
How did it perform?

Viewing held up, even when registration didn’t.

  • 783,992 visitors and 225,210 registrations

  • 103,815 live views and 33,880 on demand

  • 17,284 clicks to buy Quest 3 and Ray-Ban Meta

Registration dipped from 2023 with no paid media, but people still came to watch.

How did I test it?

The event was clear. Taking part wasn’t.

I tested the public homepage with 13 people on dscout. They rated it clear on what Connect is (4.2 out of 5) and how to engage (4.4), but:

  • Nine expected to watch online and no one expected to attend only in person, even though the page listed Menlo Park.

  • Most looked under “Program” for the schedule but weren’t sure it was there.

  • Register buttons came before the agenda and speakers.

Diagram of how the synthetic panel was built: 400+ real people applied, 135 were screened, and 100 digital twins were built 1:1 from real interviews.
What did I recommend?

Orient people before asking them to register.

  • Say who can attend in person and who should stream, right at the top

  • Put the agenda, schedule and speakers up front

  • Use fewer Register buttons and more paths to real information

Where does it fall short?

I was upfront about its limits.

  • 13 people is directional, not proof.

  • The metrics describe, they don’t prove. Registration dipped alongside a cut in paid media.

What came of it?

Outcome

  • Shipped Connect in 2023 and 2024

  • 783K visitors, 225K registrations and 17K product clicks

  • My framework reused for Connect 2025 and 2026

  • Fixes for the next site, grounded in testing