THE WORK
The brief, the work & what changed.
Six projects across marketing, product, help content, content systems and AI. Each one shows the brief, my role and what changed.
CAMPAIGN STRATEGY
THOUGHT LEADERSHIP
B2B MARKETING
ATB VENTURES · OLIU DIGITAL IDENTITY
Helping a new digital identity platform find its buyers
Oliu sold identity verification to governments, insurers and enterprises, so every piece had to make a technical product feel trustworthy and clear. I drove the strategy behind their web presence, including their website, marketing and social campaigns, thought leadership blogs and whitepapers, top and bottom of funnel email campaigns, and ongoing content governance.
My role: Wrote the content across four campaigns from 2023 to 2024, from thought leadership to a product launch, and audited the platform’s existing content. Here are some samples of my work.
149
leads from the digital identity campaign
3
sales opportunities, from 8 meetings
48%
email open rate, nearly double the benchmark
4X
downloads from one call-to-action change
Insurance campaign - Ad 1
Impressions: 23,251
Clicks: 198
CTR: 0.85%
Insurance campaign - Ad 2
Impressions: 27,859
Clicks: 223
CTR: 0.80%
Sample LinkedIn posts
CONTENT STRATEGY
INFORMATION ARCHITECTURE
UX WRITING
CONVERSION COPY
H&R Block Canada · DIY tax software · support.hrblock.ca
Making doing your own taxes feel doable
More than a million Canadians file through it in a four-month rush, so every screen, help article, and every offer had to work the first time, in English and French. I led UX writing and digital content for the DIY tax software, and grew the practice to three UX writers, two technical writers and two translators. A few of the pieces I worked on are below.
My role: Led UX writing and digital content. I owned the help centre end to end, wrote the Expert Review copy and ran its A/B tests, and grew the practice to three UX writers, two technical writers and two translators.
70%
fewer calls to the Support Centre
49%
higher attach rate on the upsell
50%
more revenue from the upsell
$69K
in team efficiency savings
HELP CENTRE
EXPERT UPSELL
SIMPLIFIED TAX FORMS
support.hrblock.ca, organized around who people are and what they came to do, with thousands of English and French articles.
A redesigned Expert Review offer, A/B tested against the old page to find what made people say yes.
Plain-language labels, instructions and error messages that made complex tax forms easier to fill in on your own.
“Pretty straightforward. And if you aren’t sure about something, the help button is actually very resourceful.” — Customer NPS comment
CONTENT AUDIT
WEB CONVERSION COPY
INFORMATION ARCHITECTURE
SEO
TELUS Digital · telus.com
The impossible audit
TV and internet were the biggest portfolio on telus.com, and I was the person looking after it. A site that size grows one update at a time, so before changing anything, I needed to know what was really there. A 5,000-page content audit showed what was out of date, what repeated itself and what was missing. That became the map for the Optik TV rebrand, which I led with external agencies, and the standard I held the TV and internet pages to day to day. Alongside it, I wrote blog content built for search, and partnered with design and engineering to update a page and prove the work.
My role: Owned the TV, internet and SmartHome+ sites. I ran the audit, led the Optik TV rebrand with external agencies, and wrote the SEO blog content.
200
dead pages retired, cutting funnel drop-off and SEO churn
4.7%
more orders
13.7%
lower bounce rate
24%
fewer single-page visits
COMMS STRATEGY
CONTENT STRATEGY
ONTOLOGY
AI-ASSISTED WORKFLOWS
Thinkific · course creation platform
Building content strategy & design from scratch
I joined Thinkific to build content design from the ground up: no team and no playbook. In a product that helps creators sell their courses, words show up everywhere, from billing screens to a learner’s first lesson. So the first job was giving everyone the same language.
I wrote the content style guide and more than 13 ontologies, shared definitions that product teams, engineers and their AI tools could all rely on. Then I put that foundation to work in three places where customers were getting stuck.
My role: Started as the first content designer, then became Senior Manager of Content Strategy & Design and led a small team. I wrote the style guide and ontologies, the MFA comms strategy, the permissions taxonomy and the import flow copy, and I swapped ad hoc meetings for office hours, saving 25 meeting hours a week.
27X
daily adoption of a security feature
+0.8
points on ease of use for permissions, shipped in under 5 days
100%
import completion, up from 94%
13+
ontologies for teams and their AI agents
CONTENT SYSTEMS
DOCUMENTATION
GOVERNANCE
AI
Thinkific · Chalk design system
Content guidelines that work without me in the room
Chalk, Thinkific’s design system, had components but no rules for the words inside them. In two months, I wrote content guidelines for more than 50 components, written for both people and AI: scope, decision rules, what never to do, accessibility and localization. Then I put them where the work happens: skills that fetch the live guidelines, and a copy check that runs on every pull request.
My role: Wrote the component guidelines and both Claude skill prompts, and helped define the pull request copy check.
50+
components with content guidelines
80%
of flagged copy issues fixed before shipping
18
pull requests moved from fail to pass on quality scores
Got a content problem?
Let's chat.
WORKING WITH AI
PROMPT DESIGN
AI EVALUATION
THINKIFIC · THINKER, A LEARNER-FACING AI BOT
Teaching an AI bot what “good” sounds like
Thinker was Thinkific’s first AI bot built for learners: the people taking courses, not the creators selling them. A bot like that talks to thousands of people you’ll never meet, so “pretty good” isn’t good enough.
I wrote the prompts that shape how it responds and how it reasons. Then, because you can’t improve what you can’t measure, I built the rubric we used to grade it. It started with three criteria: clarity, consistency and usefulness. It grew to more than ten, adding accuracy, tone, empathy, sensitivity and more, each with four scoring levels and its own evaluator prompt. I also co-built the golden dataset of model answers we tested it against.
That turned quality from a gut feeling into something the team could see, discuss and improve.
My role: Wrote the response and reason prompts, built the evaluation rubric, and co-built the golden dataset.
50→63%
of conversations graded “Good,” over one quarter
10+
evaluation criteria, up from 3
4
scoring levels for every criterion