Scale Beyond the One-to-One
If your service requires a human on the other end of every interaction, it can never serve everyone.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 85%
Wojcicki calls this one of her most fundamental principles: if you want to improve healthcare for everyone, it cannot depend on a one-to-one interaction with a clinician. She reaches for the hotel analogy — booking a room used to mean waking up in the middle of the night to phone Paris, and now nobody under a certain age has ever called a hotel. Healthcare must undergo the same transformation, delivering care for many people with the oversight of just one. Crucially this is not about deleting the expert. Genetic counselors are in permanent shortage; you cannot train every clinician; so the model becomes self-serve access with expert oversight and human capacity reserved for judgment and empathy. She sees AI accelerating exactly this shift.
Origin
23andMe's central conflict with the medical establishment was whether people could receive genetic health information without clinical supervision. Wojcicki's argument — proven over seven years of FDA submissions — was that people can spit in a tube and understand the results themselves, and that insisting on a clinician in the loop caps healthcare at the number of clinicians.
Core principles
- 01Any service gated by a one-to-one expert interaction is capped by the supply of experts.
- 02The goal is care for many with the oversight of one, not the removal of the expert.
- 03Every industry that scaled did so by making the default interaction self-serve and the human exceptional.
- 04Expert shortage is a permanent condition — designing around it beats waiting for it to be solved.
How to run it
- 1
Find every one-to-one gate
Map each point where your service requires a specific expert to be present for a specific customer. In healthcare, even a simple iron blood test requires a prescription from your doctor.
- 2
Separate judgment from gatekeeping
For each gate, ask whether the expert is exercising genuine judgment or simply holding a key. Gates that exist from habit or control are the ones to redesign first.
Pro tip Ask what would break if the gate vanished — if nobody can answer concretely, it's gatekeeping.
Watch out Removing a gate that carries real judgment is how you get a warning letter.
- 3
Redesign for oversight, not presence
Rebuild the flow so one expert oversees many customers rather than accompanying each one. 23andMe's model is self-serve results with a clinical team available on demand.
- 4
Prove quality survives
Generate data showing outcomes hold without the one-to-one. 23andMe spent years proving that people can spit on their own and understand medically meaningful results without supervision.
Watch out In regulated industries this proof is mandatory and slow — 23andMe took roughly seven years.
- 5
Reinvest the freed human capacity
Point the expert time at what only humans do — hard cases, coaching, empathy. Wojcicki expects AI to make physicians truly great at taking care of people, since AI does not get tired or have bad days.
In the wild
Wojcicki describes waking up in the middle of the night to call a hotel in Paris to book a room — a one-to-one interaction that most listeners have never experienced. Expedia and its peers made booking self-serve and therefore infinitely scalable.
→ The analogy anchors her argument that healthcare must undergo the same shift from one-to-one to scalable, and that today's insistence on a clinician per interaction will look as absurd as calling a hotel.
Recognizing that customers who learned their chronic kidney disease or type 2 diabetes risk hit a wall with no one to guide them, 23andMe acquired Lemonaid Health in 2021 to bring clinicians, telemedicine, and a pharmacy in-house.
→ Customers gained direct genetic access plus an on-demand expert team to coach behavior change — self-serve data with expert oversight rather than expert presence.
Common mistakes
Reading 'scale' as 'remove the expert'
Wojcicki's formulation is care for many with the oversight of one. 23andMe added clinicians via Lemonaid precisely because self-serve data alone left customers hitting a wall.
Assuming the expert shortage will be solved
There has always been a shortage of genetic counselors and you cannot train every clinician. Designing around the bottleneck beats waiting for the pipeline to fix itself.
Removing the gate before proving safety
In regulated industries, scaling past the one-to-one without the data to defend it invites exactly the shutdown 23andMe received in 2013.
Is it for you?
Best for
Founders and operators in expert-gated service industries — healthcare, legal, financial advice, education — trying to expand access without collapsing quality.
Not ideal for
Services where the human relationship is the product itself, or high-stakes judgment calls where no dataset yet supports removing oversight.
From the transcript
“One of the most important things for me as a takeaway from my Healthcare days if you want to improve health care for everyone it…”
“You are surprised now if you ever have to call a hotel to book a reservation... now you just go online and you book it…”
“Physicians are people like they get tired they have bad days like I think you're going to be able to make Physicians truly being great…”
From the episode
Anne Wojcicki: How 23andMe is Disrupting the Healthcare Industry
Anne Wojcicki