I design content.
I build tools.
I measure both.

The problem

Cvent's Meeting Management platform split its workflow components across separate, siloed pages. Admins setting up Auto Event Creation, Budget Policies, and other automations couldn't see how those pieces connected. They'd change one setting and break something else, then file a support ticket when they couldn't figure out why.

• No visual map of component relationships or dependencies
• Labels like "Triggers" meant nothing to users. The terminology was engineering shorthand, not user language.
• Error states gave no guidance on what to fix or what would break downstream

What I did

I introduced a centralised Workflow Hub, consolidating all components under one interface. Renamed "Triggers" to "Workflow Automations" and wrote consistent labels and tooltips that explained each component's role in context.

I created a Workflow Map showing how settings in one area affected others. This let users make changes with confidence instead of guessing.

Content was A/B tested via Maze: two versions with different label language and step sequencing. The version with better task completion and navigation scores shipped.

Figma Variables managed content states across light/dark mode and active/disabled states. What used to take hours across dozens of screens took seconds, and dev handoff was cleaner because tokens were standardised.

Copy examples

The problem

Same component, five different writers, five different choices. One tooltip said "This helps you..." Another said "Use this to..." A third skipped the explanation entirely. No one was wrong. There just wasn't a standard.

What I built

A component-by-component style guide covering every major UI pattern: accordion, avatar, badge, button, card, checkbox, date picker, dropdown, flyout, form, icon, modal, multi-select, navigation, pagination, progress indicator, slider, table, textbox, and more.

For each component:

• Tone: which register fits (e.g. friendly for empty states, direct for errors)
• Character and word limits, so labels don't break layouts
• Do's and Don'ts with concrete examples, not vague principles
• A11Y: ARIA labels, alt text patterns, what screen readers need
• L10N: what breaks in translation, symbols to avoid, text expansion allowances
• CTA standards: action-verb-first, character limits, no "Click here"

The problem

Content decisions were made by whoever spoke loudest in the room. No data, no framework, no way to prioritise. I wanted content strategy at Cvent to be defensible, backed by numbers, not preference.

The approach

Two audit modes: a full audit for a complete content inventory, and a rolling audit for ongoing quality maintenance. Standards didn't slip between reviews.

Flesch-Kincaid scoring sat at the centre. Every piece of product content was scored against the FK readability formula. Target range was 60-80 (7th-9th grade), readable for global enterprise users without being oversimplified.

The 10-dimension scoring model covered: strategy alignment, usefulness, accuracy, clarity, accessibility, readability, inclusivity, findability, engagement, and action quality.

The framework became repeatable. New content gets audited against the same standard, making quality measurable over time rather than a subjective call.

The problem

Across Cvent's product, sentences regularly started with "Because," "While," "Although," and "Once." These constructions bury the action. Users scanning a notification or error message miss the thing they actually need to do.

What changed

Updated the style guide with a clear rule and restructuring alternatives. Ran workshops where writers rewrote real examples in real time. Audited existing product content and revised alongside the team.

Real examples from the product

Impact

• Faster scannability. The outcome appears before the condition.
• Shorter strings improve localization accuracy and speed
• More consistent, direct brand voice across all surfaces

The problem

Content design was always downstream. By the time the copy request arrived, the IA was locked, the flows were built, and the only option was to write around a broken structure. I wanted to change when content designers entered the process.

What I did

Discovery: talked to engineering, product, support, and marketing to map every existing touchpoint and every gap where content design wasn't involved but should have been.

From there I built a documented collaboration model showing when and how content designers engage across the product lifecycle: ideation, design, build, ship, post-launch.

Implementation ran as structured 1-hour sessions with stakeholders. Each covered role overviews, touchpoint mapping, open Q&A, and agreed next steps.

Result

• Earlier content involvement meant fewer structural rewrites late in the sprint
• Teams outside UX understood what content design actually does and when to bring us in
• The framework scaled as new teams joined without needing to be rebuilt each time

The context

Birdeye's platform spans reviews, listings, surveys, social media, and messaging. Getting consistent content across all of those, written by different designers across different sprints, required a system, not a document.

What I built

• Error message frameworks: empathetic, actionable patterns for validation errors, permission errors, network failures, and empty states, with templates for each
• Messaging patterns: reusable templates for confirmations, onboarding nudges, tooltips, and notifications
• Voice and tone guidance: context-sensitive rules for different surfaces and user emotional states
• AI agent frameworks: goals, triggers, tasks, and output standards for Birdeye's AI automation features, so automated actions feel transparent rather than robotic

The system uses AI to surface guidance contextually. Writers get relevant standards in context, not buried in a document they have to search through.

The problem

Phoenix was a full redesign of Cvent's venue sourcing experience. More features, a different UI structure, and five completely different user types, each migrating from a different starting point. User feedback via Pendo told the story directly:

Five different users, five different approaches

• First-time free planners: full walkthrough from scratch
• Existing free planners (migrated): quick-start tooltips showing what moved where
• New enterprise planners: feature overviews based on access level
• Existing enterprise planners (Phoenix default): migration-focused orientation
• Existing enterprise planners (Phoenix not default): conversion-focused nudges toward adopting Phoenix

How I built it

All onboarding ran through Pendo. No engineering tickets, no sprint dependency. Interactive tooltips, step-by-step walkthroughs, and in-app training videos at key decision points. Content covered the full sourcing journey: RFP creation, venue shortlisting, bid comparison, post-award contract management, and budget tracking.

The problem

"SF," "sq ft," "ft2": three abbreviations for the same thing, in the same product, sometimes on the same page. International users had no metric option. Layout-specific space requirements were unexplained, which led to consistent calculation errors.

Research and decisions

"sq. ft." was chosen over "SF" (informal, unclear) and "ft2" (too technical). Used consistently across all UI elements, tooltips, and documentation.

A toggle was introduced for switching between sq. ft. and sq. m, with location-based defaults. One detail worth noting: the Metric system doesn't use periods in abbreviations, so "sq. m" not "sq. m." A standard Cvent wasn't following before this work.

Layout-specific tooltips were added to explain why space requirements vary by seating type.

The context

Pharma event managers track healthcare professional (HCP) expenses across events for regulatory compliance. The SPP (Spend Per Person) feature in Cvent's budget tool was supposed to handle this, but inconsistent labels, unclear admin setup, and no cross-event reporting made it unreliable for compliance use.

What I did

Terminology standardisation: "Spend Per Person" and "SPP" were used inconsistently across labels, descriptions, headers, and config screens. I mapped every instance and established a single consistent standard.

The setup page for SPP configuration was confusing two distinct workflows. I defined and rewrote them clearly:

Community guidance articles were written so pharma clients could answer their own SPP questions, reducing support load and giving compliance teams a reference they could point to.

The lesson

In regulated industries, inconsistent terminology isn't just confusing. It creates compliance risk. This project aligned with Cvent's 2024 customer-centricity goal: making the product work better for the users who depend on it most.

The problem

Writers were writing for products they'd never clicked through end-to-end. Localization teams were translating strings without understanding the workflow they lived in. Quality suffered because context was missing.

The format

• 10-15 min: Workflow overview, what it does, who uses it, what matters
• 30 min: Hands-on walkthrough of the actual product, not slides, the real interface
• 15 min: Open feedback, what's unclear in the flow, what's off in the copy

Sessions ran on a regular schedule, attended by both teams. No slides, no theory. Just clicking through the product together and talking through what we found.

What it changed

• Writers made fewer assumptions about how features actually work
• Localization teams could flag translation risks earlier, before strings were locked
• Both teams built shared context that made collaboration faster in future sprints