Disease Intel Canvas
A modular, compliance-structured launch intelligence system
Overview
Supported 5 concurrent launches, <8-week to launch a new model, and 41% lower cost per identified patient with an upstream planning and optimization system built for lean rare-disease economics
Instead of rebuilding the same launch work disease by disease, the system created reusable strategy, content, and scoring layers that preserved economics on lean rare-disease programs.
Goal
Help a 7-person team generate launch-ready disease intelligence and optimization guidance without rebuilding workflows per disease—while preserving economics on programs starting at ~$30K/month, where traditional agency overhead was never viable.
Team
Led product, strategy / activation, media / content, analytics / optimization, and collaborators supporting audience definition and deployment.
Role
Created the system concept, sequenced the build by business impact—cohort automation first, keyword discovery deferred—and operationalized the highest-leverage components: cohort generation, creative orchestration, and campaign scoring.
Problem
Supporting 5 concurrent rare-disease launches on budgets starting around ~$25K/month would traditionally require ~842 hours / >$140K of external support per program
The challenge was not just research volume—it was specialization under low-prevalence, regulated-market conditions.
These launches spanned rare neurologic, nephrology, hematology, oncology, and inherited skeletal disorders across populations of ~1,200–16,500, and each required niche disease understanding, compliant content, and different audience logic.
Because programs often started with only ~$25K/month in working media, traditional agency-style planning overhead was economically mismatched from the start.
Traditional support economics did not fit rare-disease budgets
- ~$25K/month working media was a common starting point in early programs
- Estimated annual external support for one program alone: 842 hours / $139K
- Channel-specific support estimates included 156 hours / $25K SEO, 240 hours / $38K paid search, 225 hours / $36K paid social, and 221 hours / $40K analytics
Low prevalence increased the need for leverage
- Patient Activate programs spanned populations of ~1,200–16,500
- The platform was deployed across 5 rare models, each with different burden, clinical language, and care context
- Limited market size made planning efficiency economically critical
Each disease required different audience logic
- Launches spanned neurologic, nephrology, hematology, oncology, and inherited skeletal disorders
- Inheritance pattern, age of onset, and care context could shift the primary audience from patient to parent, adult child, or family unit
Regulatory and channel rules slowed every workflow
- FDA / PRC review constrained copy, visuals, and educational framing
- Unbranded disease-awareness rules and media specs varied by channel
- Every output had to be compliant enough for review and structured enough for reuse
Solution
Turned rare-disease launch prep into a modular Plan → Develop → Manage system that supported 5 launches and <8-week setup
We initially scoped audience intelligence and keyword discovery together. When pressure-testing showed the audience-planning layer had the most immediate ROI—and deeper keyword automation required more program maturity—we narrowed scope to keep the system lean enough to ship across five concurrent programs.
Strategy / Activation
Audience intelligence was sequenced first—delivering more immediate ROI than deeper keyword automation.
- Faster disease understanding
- Search demand + whitespace
- Audience hypotheses and launch strategy
Media / Content
Raw generative image outputs required Topaz AI enhancement to reach compliance-ready quality—a workflow dependency flagged for the next design iteration.
- Faster brief development
- Keywords, copy, visuals
- Launch-ready compliant inputs
Analytics / Optimization
Daily Zoho→BigQuery reporting was abandoned—data volume limits and cost made it non-viable. Monthly staged transforms replaced it.
- Qualified-patient learning
- Better future planning
- Scoring and refinement inputs
Scale
I sequenced the build by business impact—audience intelligence first, creative orchestration second, deeper keyword automation deferred—reducing hours of net-new work per launch.
Scale came from sequencing decisions, not headcount. Instead of rebuilding the same launch work disease by disease, the system created reusable strategy, content, and scoring layers so a team of 7 could support 5 concurrent launches and avoid the ~$139K per-program overhead a traditional model would have required.
Research + Search
Operationalized
- Agentic audience intelligence workflow
- Keyword intelligence + classification
Why it mattered
- Reduced repeated disease landscaping and search-strategy rebuild across launches.
Content + Creative
Operationalized
- AI image workflow orchestration
- Reusable copy + content inputs
Why it mattered
- Compressed creative production burden by an estimated ~500 hours across landing pages, ads, HTML5, motion, and revisions.
Feedback + Optimization
Operationalized
- Campaign scoring loop
- Qualified-patient keyword refinement
Why it mattered
- Made future launches smarter by turning live campaign results into reusable scoring and planning inputs.
Intelligence
The learning loop cut cost per identified patient 41% while increasing diagnosed patients >3x and HCP handoffs >7x
I built the scoring mechanism so campaign results could feed back into keyword classification and audience refinement—making disease intelligence cumulative. Our first pipeline tried combining 4 data sources in a single query. Elegant in design, too fragile and resource-intensive to run reliably. We rebuilt as staged transforms with independent failure modes. Less clever—actually reliable enough to run month after month.
Learning Inputs
- Campaign results
- CTR / CPC / CPA trends
- Qualified-patient rates
- Engagement and conversion patterns
- Emerging keyword and theme performance
- Segment-level response signals
System Improvements
- Keyword scoring
- Keyword classifications
- Priority search themes / territories
- Segment-mapped planning inputs
- Content optimization recommendations
- Future launch strategy
Why It Mattered
- Made disease intelligence cumulative instead of one-time
- Helped identify higher-yield search themes earlier
- Improved future launch planning with real program evidence
- Created the evidence base that reduced cost per identified patient 41%—not by spending less, but by spending on the signals that actually converted.
The value was not just better reporting—it was a reusable learning loop that improved strategy, content, and targeting with each launch.
Artifacts
The Disease Intel Canvas materialized as a working system for launch planning, execution, and learning
Audience design, keyword intelligence, workflow orchestration, and program monitoring were surfaced through one modular interface.
Audience Intelligence
Keyword Intelligence
Content Orchestration
Program Monitoring & Learning Loop
The system translated disease intelligence into launch-ready creative, messaging, and channel-specific outputs
Shared intelligence layers produced tailored campaign assets shaped by disease burden, audience context, and platform requirements.
