
Data Wizard
Raise a hand and a “Power Ball” of data snaps into your palm; open your mouth to feed it until it explodes into animated data — with easter eggs triggered by blinks.
Case Study
Three branded AR filters + a physical selfie wall, driving user-generated content across a 20,000-attendee conference.
Splunk .conf19 · #SplunkAR
Executive Producer — Jill Honig, Awesome Girl Productions
Splunk wanted attendees at .conf19 to become the brand — not just watch it. The answer: a suite of three augmented-reality filters on Facebook's Spark AR platform, engineered to generate a flood of shareable, hashtagged #SplunkAR content the on-site social team could pull straight onto the event's digital signage.
I executive produced the program end to end — owning the team and the delivery — coordinating a specialized AR development studio, the client's brand and social teams, and Facebook's platform-approval pipeline.

Each filter turned a Splunk brand pillar into an interactive, gesture-driven experience.

Raise a hand and a “Power Ball” of data snaps into your palm; open your mouth to feed it until it explodes into animated data — with easter eggs triggered by blinks.

A full-CGI head replacement that turns you into Splunk's mascot — mimicking expressions in real time with eyebrow, mouth and head-shake triggers.

A poster generator outfitting you in a CGI fez and cape, with animated Splunk-branded backgrounds and rotating quote overlays.
Anchored on the show floor by a 12′W × 8′H AR wall — QR codes and instructions on the front, a geometric selfie backdrop on the back.
Facebook AR filters don't just need client approval — they have to clear Facebook's own platform review before going live, and that review is a black box you can't rush. An already-tight creative timeline became a hard, externally governed deadline with zero slack.
Miss the Tuesday client approval by even a few hours, and the filters risk not clearing review before doors open — at which point the activation simply doesn't exist. There was no version of “we'll fix it on site.”
Buttercup required a full CGI head tracking expressions from every angle — sophisticated real-time 3D. I directed the studio's sprints against the approval deadline, keeping build, feedback and revision in parallel, not in sequence.
With a Tuesday lock there was no room for open-ended loops — each round went in as one batched, documented revision (Data Wizard's blink colour shifts, bright-pink eyes, sparkle particles), not a trickle of one-offs.
The schedule was built backward from Facebook's review window, not the client's approval date — “client-approved” and “live for attendees” treated as two deadlines with the black-box review protected between them.
The go-live playbook shipped with the filters: geo-targeted Spark AR “Try” posts boosted to the conference location, one per filter, Monday through Thursday — so the filters didn't just exist, they got discovered.
Approved, cleared, and live for doors.
All three filters cleared client and Facebook platform approval on schedule and went live for .conf19 — turning attendees into Data Wizards, Buttercups, and Trust members across Facebook and Instagram, and feeding a stream of #SplunkAR user-generated content onto the conference's digital signage.
Watch the work
AR Filter videos — playlist

The Takeaway
When delivery depends on a platform gate you don't control, build the schedule around the gate — not around your own team's deadlines.
Protect the window you can't see into, and everything upstream falls into line.