Why decentralised trials now produce better data than the lab
The research model you choose is a commercial decision before it’s a scientific one. Here’s why you should care about the difference.
There are two things I want to cover in this piece, because they’re the two points I hear most often from founders weighing up decentralised vs lab-based research.
The first is data quality.
Most founders assume, by default, that lab-based research produces better data. For the majority of questions a supplement or nutrition brand actually needs to answer, that assumption is wrong.
Decentralised trials now produce better data than the lab. Not equivalent. Better.
The second is practicality and speed.
This usually gets filed under “operations”, a nice to have, but secondary to the science.
That’s a mistake.
Speed, recruitment reach, and iteration velocity aren’t logistical conveniences. They compound into a commercial moat that slow, lab-based research simply can’t produce.
Let’s take them one at a time.
Part 1: The data quality question
The lab used to be the right answer. Then the tech caught up.
Five or ten years ago, anyone arguing for decentralised trials over lab-based ones would have been wrong. The tools to capture meaningful data from a participant’s kitchen didn’t exist at scale.
That’s no longer true. What’s now available at a participant’s front door:
Wearables tracking sleep architecture, HRV, and recovery every night for months
Finger-prick blood kits returning validated lipid, glucose, and inflammatory panels by post
Bluetooth body composition scales producing a reading every morning
At-home gut microbiome kits shipped direct
Continuous glucose monitors giving minute-by-minute metabolic data
This is the infrastructure shift that makes the rest of the argument possible.
Where the lab quietly loses
If you’re doing acute mechanistic work, stable isotope tracers, muscle biopsies, tightly controlled metabolic physiology, the lab is irreplaceable.
No home-based alternative exists for a biopsy needle.
This is exactly what my own PhD is built on, so I’d know.
But that’s maybe 10% of what brands actually want to know. For the other 90%, sleep, body composition, recovery, gut health, cognitive function, mood, the lab is no longer the gold standard.
Here’s why:
Two data points vs eighty-four
A classic lab RCT measures the primary outcome twice: pre and post.
Body composition scanned once at week zero, once at week twelve.
Sleep captured via questionnaire at two time points.
The difference between those two numbers gets called “the effect.”
The problem: a single body composition scan has enough measurement noise to swamp most realistic supplement effect sizes.
Hydration, meal timing, day-to-day bodyweight fluctuation.
A glass of water before you step on the scale moves the number more than three months of your supplement will.
You’re measuring noise and calling it signal.
Here’s what this looks like in practice.
Imagine a sleep supplement brand running a standard lab trial: 40 participants, one visit at baseline, one at week 12, with sleep measured via the Pittsburgh Sleep Quality Index at each end. Two years to recruit, run, and analyse.
Outcome: a 0.3-point shift on a 21-point scale. Statistically significant? Maybe.
Meaningful to a customer?
Not really.
Run the same product decentralised with 60 participants wearing a sleep tracker every night for 12 weeks. You get 5,040 participant-nights of sleep data instead of 80 questionnaire snapshots. The noise averages out. The signal tightens.
You see a 14% improvement in deep sleep duration, a number your customer can see on their own wrist, and you’d have missed with the lab design.
Same product. Same duration. Completely different evidence base.
The “objective” lab measures aren’t that objective
A lot of what makes lab studies look rigorous is questionnaire-based:
Sleep → Pittsburgh Sleep Quality Index (self-reported recall)
Diet → three-day food recall (self-reported, notoriously inaccurate)
Digestive symptoms → symptom diaries
These are the same self-report instruments a decentralised study would use, except the lab version loses the continuous objective layer (wearable, CGM, scale, blood panel) that modern tech makes possible at home.
Ecological validity: the customer doesn’t live in your lab
A lab participant knows they’re in a study. They eat the food the metabolic kitchen gives them, sleep in an unfamiliar bed, and behave like someone being watched.
Your customer doesn’t.
If your product works in a participant’s actual life, their kitchen, their bed, their routine, you’ve learned something real.
If it only works in a windowless basement, you’ve proven very little about what happens when someone actually buys it.
“But how do you know they actually took it?”
This is the first objection every founder raises, and it’s a fair one. In a lab, someone watches the participant take the dose. At home, they could leave the bottle in the cupboard. So surely lab wins on compliance?
In practice, the opposite is usually true. Modern decentralised trials have several compliance advantages the lab can’t match:
Timestamped dosing logs via app-based check-ins every day, not a single study visit every few weeks.
Device data as a compliance signal. If a participant’s wearable shows they took the product as instructed, you’ve triangulated adherence from an independent source.
Participants are in their normal environment, so dosing fits into their routine rather than being disrupted by clinic visits.
The old mental model: “lab = watched, home = trust me” is from 2015.
Today, a well-designed decentralised trial captures far more compliance data points than a monthly in-person visit ever could.
Part 2: The practicality and speed question
Founders hear “decentralised = faster” and file it under operations. That’s a mistake. Speed is one of the most commercially important things your research strategy produces.
Speed kills — and slow research kills brands
You need data yesterday, not in three years.
A competitor shipping a claim-backed product in twelve months while you wait on a university slot that opens in Q3 2027 is a competitor eating your market.
Iteration velocity compounds
If you can run five well-designed decentralised trials in the time it takes to run one lab-based study, you’re not just doing more research.
You’re building a proprietary evidence base faster than anyone can copy.
One big university trial gets you a paper.
Five trials gets you a moat.
IP and trust compound with every trial you ship, and that compounding is the part your competitor can’t shortcut by writing a bigger cheque.
Bigger, more representative samples
This is where speed and data quality converge. Lab trials recruit from one city, usually within driving distance of a single university. That means your sample is biased toward whoever lives near that university, often students, often one demographic slice.
Decentralised trials recruit nationally.
You get a sample that looks more like your actual customer base, and you get it faster.
More participants and more measurements per participant. Statisticians will rightly point out that a bigger N doesn’t automatically fix biased measurement, true.
But a representative national sample, combined with continuous objective data, beats a small convenience sample with two measurement points on both counts.
The single-lab bottleneck
Lab studies run at the speed of one researcher bringing one participant into one lab at a time. That’s not a queue, it’s a structural cap.
60 participants × one coordinator × one baseline visit each = weeks of onboarding before the intervention even starts. Then weeks more at the end for follow-up. Decentralised trials run participants in parallel. People onboard asynchronously, from anywhere, with no lab slot to book.
The same study that takes 18 months in a lab can run in 6 months decentralised not because the science is shortcut, but because the bottleneck is removed.
Tangible outputs your customer understands
And then there’s the piece that actually sells product.
A statistically significant shift on a clinician-administered Likert scale means nothing to the person buying your sleep stack.
A 14% improvement in their Oura sleep score?
That’s tangible.
That’s shareable.
That’s a claim that translates into behaviour change, and repeat purchases.
Decentralised research generates the metrics your customer already cares about, because they’re the metrics your customer is already tracking on their own wrist.
The bottom line
The useful question isn’t “lab or decentralised?” It’s: what’s the fastest route to high-quality data that holds up and that your customer can actually feel?
For acute mechanistic work, still the lab.
For basically everything a supplement or nutrition brand needs to substantiate a structure-function claim: sleep, body composition, gut health, recovery, cognitive function, wellbeing, the answer has moved.
And for these claims, regulators and retailers are not asking for a university-stamped RCT. They’re asking for rigorous evidence. Decentralised trials produce that.
The brands figuring this out first are going to ship better products faster, build a compounding evidence base, and run circles around the ones still waiting on a university slot that opens in 2027.
If you’re thinking about what an evidence roadmap looks like for your brand, chat with me.
— Nathan