Partner, Not Human Subject
People will give you their most valuable data — if you treat them as collaborators instead of specimens.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 85%
23andMe's research model rests on a reframing: people are not unwilling to share data, they are unwilling to be human subjects. Wojcicki argues traditional healthcare research treats you as someone who won't come back and won't be responsible, so it extracts as much as it can — an experience no consumer company would design. 23andMe instead makes research a partnership: an intake survey on joining, surveys about the diseases each person already cares about, and results that flow back to the individual. Everyone has a cause — Parkinson's, Crohn's — and everyone can be a control for someone else, so no participant is useless. Participation is opt-in and never defaulted, and over 80 percent of customers opt in.
Origin
Wojcicki was inspired by the HIV activist movement and organizations like Susan G. Komen and Livestrong, where people participated in disease research with real enthusiasm. Her question was how to capture that enthusiasm while making it far more data-focused, and to fix the fragmentation of research trapped in small institutional cohorts at Stanford, Harvard, and the NIH.
Core principles
- 01The objection 'people won't share their data' is usually a symptom of a bad experience, not a real preference.
- 02Everyone has a disease or cause they care about — that motivation is the recruitment engine.
- 03Participants must be able to be controls for each other; every individual is useful to the pursuit.
- 04Opt-in must be genuine and never defaulted, or the partnership is a fiction.
How to run it
- 1
Diagnose the real objection
Test whether people refuse to share data because of privacy per se or because of how they are treated. Wojcicki's read is that healthcare treats people so poorly that refusal is a response to the experience.
- 2
Give something back first
Lead with value to the individual — their ancestry, their traits, their health risks — before asking them to contribute to the collective. The personal result is what earns the right to ask.
- 3
Connect to the cause they already have
Surface the condition each person cares about via an intake survey. One person cares about Parkinson's because of family, another about Crohn's — the motivation already exists.
Pro tip Frame contribution as 'we're all in this together, we're going to find the cures together' rather than as a data request.
- 4
Make every participant useful
Design so people with no relevant condition serve as controls for those who do. This means nobody is turned away and everyone's contribution has a real role.
- 5
Keep opt-in genuine
Never default participation on. 23andMe's research participation is opt-in and never defaulted — and over 80 percent of customers opt in anyway, which is the proof the model works.
Watch out A defaulted opt-in destroys the partnership framing and turns you back into an extractor.
In the wild
Against the prevailing argument that people didn't want their data shared, 23andMe made research participation opt-in, never defaulted, and gave customers meaningful personal results plus surveys about the diseases they already cared about.
→ Over 80 percent of customers opt in to research, and the database exceeded 14 million customers — disproving the claim that people won't share genetic data.
Wojcicki was moved by how HIV activists shaped drug discovery, trial design, and enrollment, and by the enthusiasm around Susan G. Komen and Livestrong. She asked how to capture that energy in a more data-focused way.
→ The insight became 23andMe's research arm — a crowdsourced model where millions of people participate rather than the small institutional cohorts that had fragmented healthcare research.
Common mistakes
Reading refusal as a privacy preference
Wojcicki's data suggests people refuse because of a degrading experience, not an inherent objection to sharing. Treating it as a privacy problem leads to legal fixes when the actual fix is product design.
Defaulting people into participation
Once participation is defaulted on, the partnership is a fiction and you've become the extractor you were replacing. The 80 percent opt-in only means something because it was freely chosen.
Asking before giving
The partnership works because customers receive personally meaningful results first. Requesting contribution before delivering individual value is just the human-subject model with better branding.
Is it for you?
Best for
Companies building research, data, or community platforms where participation is voluntary and the data is personal or sensitive.
Not ideal for
Contexts where you cannot give participants meaningful value back, or where the data use genuinely cannot be explained honestly.
From the transcript
“What I think we've proven out is that people are actually quite willing to participate in research they just want to be treated they want…”
“In healthcare you're just treated so poorly that people don't want to participate because you're treated like a human subject you're also treated like somebody…”
“Every individual has a disease that they care about so I might care about Parkinson's... you might care about Crohn's disease we can be controls…”
“It's something that people opt into it's never defaulted but you opt-in over 80 percent of customers opt in.”
From the episode
Anne Wojcicki: How 23andMe is Disrupting the Healthcare Industry
Anne Wojcicki