For the past decade, I’ve designed the systems, frameworks, and governance structures that make high-stakes, technically complex products legible and trustworthy to the people who use them. My work lives at the intersection of content craft, systems thinking, and regulatory judgment, across spatial computing hardware, AI diagnostic tools, and global advertising platforms that had no established design precedent.
Currently the lead content designer for first-party data management and setup in Google Ads. Previously the privacy content expert for Meta’s VR org. Newspaper reporter before all of it.
“Room scan” · “Room map”
“Spatial data”
Plain didn’t mean reassuring. My plain-language hypothesis tested worse. Research showed that “scan” read as surveillance, and the industry term tested safer, so I went with the evidence. It still ships across Meta Quest headsets several device generations later.
Additional domains detected for configuration. You may need to tag additional domains where conversion activity is detected.
Tag new domains to improve conversion measurement
We’ve detected conversion activity on domains that aren’t tagged yet. Tag them so every conversion is measured accurately.
Say what’s wrong, then what to do. One of hundreds of Google Ads diagnostics I rewrote, then encoded as rules an AI agent could apply, so the voice held at scale.
Ads said “Link” · Analytics said “Connected”
“Connected” everywhere, for the same thing
A semantic audit, not a word swap. I mapped every overlapping term across Google Ads and Analytics, then got leadership to agree on descriptive names over internal jargon, which left room for the features that came next.
The practice
Most of my work happened at the point where a product asked someone for something sensitive: their space, their location, their telemetry, their sales data. Usually there was one screen in which to explain why. I wrote for that screen, and I built the frameworks and terminology systems that kept the next hundred screens consistent.
In practice that meant treating legal, research, engineering, and product as partners rather than approvers, deprecating internal jargon in favor of words people already recognized, and using AI to generate from frameworks I defined, with human checkpoints wherever judgment lived.
Selected work
Each project starts with the constraint that shaped it, then the screens that shipped against it. Where results remain proprietary, the outcome is described rather than counted. Select any screen to enlarge.
Google tags didn’t capture the full sales picture, advertisers didn’t fully trust the data, and importing measurement data was hard enough that novice and power users both dropped off. Underneath all of it, Ads and Analytics described the same objects with different words.
I ran a semantic audit across both products, built the bridge between the two vocabularies, and defined a four-pillar content framework that the whole flow was written against.
The four-pillar content framework
Hundreds of diagnostic messages had to stay consistent, accurate, and on-brand without content design becoming the bottleneck. So I wrote the framework first, with explicit behavioral rules and logic for every message, then trained an agent on examples I authored and validated, co-created with a content design partner who owned the technical setup.
Cross-functional partners could then generate accurate copy and keep a historical record of every message, without direct content design oversight.
Input · Action · Output
| Input | Action | Output |
|---|---|---|
| An original diagnostic message | Rewrite | A message that names the problem and the fix |
| Conditions only: alert type, when it appears, when it disappears | Write | A new message built from the rules, with a title and body |
The projection that convinced product and engineering
Spatial data was a first-of-its-kind VR feature that required a net new privacy flow: what data a mixed-reality app wanted to collect, why, and what it did for the experience. No org-wide guidance existed, global regulators had requirements, and product wanted the flow as short as possible.
Research determined that an explainer screen, followed by consent, was the way through. Educate first, then ask. I named the concept against my own tested hypothesis, and specificity did the reassuring: sensors captured the size of walls, surfaces, and objects, not what was written on the paper on your desk.
Device support per Meta’s spatial data permission documentation. Last checked 21 July 2026.
As content design lead for privacy at Reality Labs, I navigated PII and regulatory constraints and influenced VP-level stakeholders to pivot a high-risk telemetry update, preventing significant legal exposure by reframing the product narrative from risk management to disclosure done properly.
The rebuild stated what was collected, how it was used, and what the person controlled, then collapsed the whole thing into a single setup moment with the context built in.
Current setting documented in the Meta Quest help centre. Last checked 21 July 2026.
A net new flow in virtual reality, asking people to turn location on at the headset level so that location-based apps could request the device’s location. People were used to location on their phone, not a VR headset. Explaining succinctly what location was used for, and how it could work, was extremely challenging, and it moved forward without time for research.
Combining an explainer screen with a system consent proved complicated, but careful collaboration with legal, product, and policy partners let me write language, aligned with regulatory requirements, that explained the risks, reasons, and specifics people needed in order to make an informed decision.
Device support and the approximate or precise choice per Meta’s VR location documentation and help centre. Last checked 21 July 2026.
As VR grew, more apps requested or gained access to more data. In the spirit of transparency, it was critical to let people know what apps were collecting, and to categorize the information so that people could understand it.
I worked closely with product and engineering to determine what elements apps could collect, what they meant, and what they were used for. Then I named the buckets each element would fit under and recommended icons that visually described each one, keeping the categories broad enough to avoid renaming as new elements were added.
As a content design leader and collaborator, I led foundational work that not only championed content design but produced scalable standards and best practices. I led critiques, reviews, and workshops, consulted on projects, and gave recommendations to cross-functional partners and content designers.
I kept in regular contact with cross-functional partners to understand goals, challenges, and opportunities, while keeping those partners informed of content design goals, best practices, and standards.
I identified a significant gap in my product team, that we couldn’t use our own product to understand user needs and pain points, and influenced product and engineering to overhaul an external-facing dashboard, improving engagement and businesses’ ability to sell across Facebook.
Sellers could see that their score was bad, but not why, or what to do about it. I standardized the scoring vocabulary, built a communications library so that every email and notification arrived with the right context, and prioritized convenience and context over aesthetic reduction.
eBay users needed a streamlined way to set up additional security features for their account. Improvements to the process and the on-product flow told users the new way they would log into their accounts, by phone call or SMS, and what to expect at each step.
The content went further to eliminate extraneous, inaccurate, and repetitive messaging, making the process easier to understand and faster to complete.
eBay’s texted-code two-step verification dates to early 2017, per an eBay statement in March 2017. Push notification verification followed in 2019 and authenticator apps in 2023. The flow is still documented in eBay’s account security help. Last checked 21 July 2026.
How I work
Before writing at volume, I write the rules: terminology, behavior, and the logic that decides what a message says. The framework is what survives after I move on.
I bring a hypothesis and I hold it loosely. “Room scan” was mine, and testing killed it. Going with the evidence is what made the naming durable.
Global regulators, legal review, shifting deadlines, someone else’s flow. I define a strict MVP, protect the V2 idea, and get the work out the door.
Experience
Google·Meta·BuildingConnected·eBay·Expedia
Contact