When I started Edison, I thought the hard part would be the science. Genomics, biomarkers, precision health - complex territory. What I actually learned is that the hardest part of building a health tech company is getting the data layer right before you build anything on top of it.
What You Need to Know
- Health tech startups face a unique challenge: the data you need is fragmented across clinical systems, labs, wearables, and patient records that were never designed to talk to each other
- Building the data foundation first, before features, is counterintuitive for startups under pressure to show product - but it's the only approach that scales
- Regulatory requirements around health data in New Zealand mean you can't move fast and break things the way consumer tech does
- The founders who succeed in health tech are the ones who understand both the clinical context and the data engineering
Starting With Why
Edison was born from a simple observation: healthcare in New Zealand is reactive. We wait for people to get sick, then we treat them. The data to predict and prevent illness already exists - in genomes, in blood panels, in family histories - but it sits in silos, inaccessible to the people who could use it.
The mission was clear. Remove the guesswork. Use genetics and data to create a highly personalised healthcare experience that catches problems before they become conditions.
The mission was clear. The execution was anything but.
The Data Problem Nobody Warns You About
Every health tech founder I've spoken to hits the same wall. You have a brilliant idea for a product. You know what the clinician needs to see. You know what the patient outcome should be. And then you discover that the data you need lives in fifteen different systems, in twelve different formats, with varying levels of completeness and accuracy.
80%
of health data is unstructured - clinical notes, lab reports, imaging - making it inaccessible to most analytics tools
Source: IBM Watson Health, 2020
At Edison, we spent our first six months on data architecture. Not features. Not user interfaces. Data models, integration protocols, and validation pipelines. Our investors were nervous. Our advisors wanted to see a product. But we'd seen what happens when you build features on a shaky data layer - you spend the next two years patching.
That decision to invest in the foundation was the single most important strategic choice we made. When we later needed to add new data sources - a new lab partner, a new biomarker panel, a new genomic assay - the foundation was ready. We integrated in weeks, not months.
What Health Tech Gets Wrong
Most health tech startups build like consumer tech startups. Ship fast, iterate, get to market. That playbook works when you're building a social app. It breaks when you're building clinical tools.
Here's why. In consumer tech, a bug is an inconvenience. In health tech, a bug can be a clinical safety issue. A data integration error that shows the wrong biomarker value isn't just a bad user experience - it could lead to a wrong clinical decision.
This doesn't mean you move slowly. It means you move carefully in the right places. Data integrity, clinical validation, and regulatory compliance need rigour from day one. The user interface, the onboarding flow, the marketing page - those can iterate.
The startups that fail in health tech aren't the ones with bad ideas. They're the ones that treat health data like any other data.
Jay Harrison
Health Technology Advisory
The Regulatory Reality
New Zealand has strong health data protections, and rightly so. The Health Information Privacy Code sets clear rules about how health information can be collected, used, stored, and shared. For a startup, this means privacy and consent aren't afterthoughts - they're architectural requirements.
We designed Edison's consent framework before we designed the product. Every data flow had a documented purpose. Every integration had a consent pathway. Every storage decision had a retention policy.
This felt slow at the time. In hindsight, it was fast. Organisations that bolt on privacy compliance later spend more time retrofitting than we spent building it in.
72%
of New Zealand consumers say they'd share health data for personalised care if they trusted the organisation
Source: Health Quality & Safety Commission New Zealand, 2020
Trust is the currency of health tech. And trust starts with how you handle data.
Lessons for Founders
If I were advising someone building a health tech startup today, I'd tell them three things.
Build the data layer first. Before your product, before your features, before your pitch deck screenshots. Get the data architecture right. Define your data models. Build your validation pipelines. Every week you invest here saves a month later.
Hire clinical and technical together. Not "build it then show a clinician." From the first week, have clinical expertise shaping the data requirements, the validation rules, and the workflow design. Clinical context isn't a nice-to-have - it's a design requirement.
Earn trust before you scale. Work with a small group of clinicians and patients. Prove the data integrity. Prove the clinical value. Get referrals based on outcomes, not marketing. Health tech scales on trust, not growth hacking.
The data-driven health startup isn't about having the most data. It's about having the right data, structured properly, governed carefully, and serving a clear clinical purpose.