Advarra Analytics
Turning clinical trial data into decisions people could actually use. AI practices across discovery, delivery, and iteration are documented below.
Product Manager · Data Platforms | Advarra
Proof in Action: Bring Analytics to the People!
The Problem
Clinical research organizations were sitting on enormous amounts of operational data, but having data wasn't the same as being able to use it.
The Product Opportunity
Make analytics accessible to the people who actually make research decisions — without requiring them to understand the underlying database.
The analytics experience was only as good as the data underneath it.
Source App → Data infrastructure → Analytics layer → Customer
How I Helped Us Get There
Defined product requirements for the analytics ecosystem.
Partnered with Engineering on data architecture and platform decisions.
Worked through the transition of data workloads toward a new standard.
Helped establish how data would be made available for downstream analytics.
Considered data quality, access, scalability, performance, and usability as part of product decisions.
Connected technical platform work back to customer-facing outcomes.
The challenge wasn't building another dashboard.
It was making analytics useful to someone who doesn't want to think about databases.
Laur’s Product Operating Model
I seek to understand the customer, their workflow, and the problem we're actually trying to solve before jumping to a solution.
Deliverable: A clear understanding of what problem matters and why.
Discover
Not every customer request, idea, or opportunity can become a product initiative. I balance customer value with business priorities, technical feasibility, dependencies, and available capacity.
Deliverable: A focused roadmap and a shared understanding of what we're doing and why.
Prioritize
Make sure the product succeeds beyond the development team.
Deliverable: Customers and internal teams know what the product does, why it matters, and how to use it.
Enable
I work closely with Engineering and cross-functional teams to translate product direction into clear, actionable work while staying engaged through delivery.
Deliverable: A product that actually makes it from an idea to a usable experience.
Deliver
Launching isn't the end of the product process. I use customer feedback, adoption, data, and what we learn during delivery to continuously refine the product.
Deliverable: The next set of product decisions.
Iterate
How I Use AI in My Product Practice
How I work with AI. I bring a working fluency to these tools, built through formal coursework and daily practice. That lets me choose the right tool for each job, write prompts that give useful context and constraints, and verify outputs against the source material before anything influences a decision. I hold AI to the same standard I hold any analysis: it has to be traceable, testable, and accountable.
Discover. I use AI to synthesize customer interviews and market research, surfacing patterns that are easy to miss under deadline pressure and turning raw signal into a clearer picture of the problem space.
Prioritize. I use AI to pressure-test assumptions before they reach a roadmap review. I treat it as my most skeptical customer, challenging my opportunity sizing and reasoning until the weak points surface. I also use it to model opportunity cost when we consider shifting priorities, so trade-offs are explicit before we commit.
Deliver. I use AI to draft acceptance criteria and edge-case lists, then review them for gaps before engineering sees them. A common outcome is catching performance issues early and addressing them before launch, when the cost of change is still low.
Enable. I use AI to build the materials that help teams and stakeholders move faster, including launch briefs, FAQs, training content, and decision logs. For the Advarra Analytics rollout, I built the adoption packet with it so teams had what they needed from day one.
Iterate. I use AI to analyze experiment results, summarize retrospectives, and turn feedback into the next set of hypotheses, so each cycle starts from what we learned rather than from memory.
Guardrails on how I use AI:
Human ownership: AI drafts and analyzes, but a person makes and owns every decision.
Verification: Outputs are checked against source material (interview notes, data, specs) before they influence a roadmap or release.
Data protection: Customer and confidential information stays within approved tools and is handled according to company policy.
Traceability: I can show where a conclusion came from, so AI-assisted work can be audited and explained.