Studio — Chapter One
Fourteen people. No juniors. No agency floor. Just a bench of strategists, data scientists, designers, and engineers who learned experimentation where it is unforgiving — and now ship that discipline into the SaaS companies that need it most.
Origin
Tyrell Lab was founded in 2019 by two experimentation leads who had spent the previous decade inside Booking.com and Stripe — running the kind of testing programs that turn a homepage into a research instrument and an onboarding flow into a sequenced set of bets. They had watched those programs compound, lift after lift, into the growth engines of two of the most consequential internet companies of the era, and they had grown tired of watching smaller SaaS teams try to reproduce that work without the bench to do it.
The thesis was simple and unfashionable. Hire only people who had already done the job inside a company with real traffic and real stakes. Cap the team at a size small enough that every partner still writes briefs by hand and reviews every readout. Refuse work where the unit economics cannot support rigorous testing. In 2021 the studio closed a $3.2M seed round from First Round Capital, almost entirely to give the founders permission to keep the team small and the standards high.
Six years later, the constraint is the brand. Tyrell Lab is fourteen people across nine countries. There is no agency floor, no bench model, no junior pyramid. Every experiment is designed by a partner with eight or more years in experimentation. That is what we sold when we were two people at a kitchen table in San Francisco, and it is what we still sell at fourteen.
The Bench
Clients do not hire individuals. They hire a four-person delivery team drawn from a single bench, with a partner signing every brief. The functions below cover the full experimentation stack — from hypothesis to shipped code to validated lift.
01
Six partners, each with eight or more years inside an in-house experimentation org. They write the test brief, sequence the roadmap, and own the readout. Every engagement has one partner on point from kickoff to hand-off — no account managers, no re-briefing.
6 partners · average 11 yrs in experimentation
02
Three senior data scientists who design the measurement plan, audit the instrumentation, run the Bayesian and frequentist analysis, and pressure-test every shipped lift. They maintain the studio's internal sequential-testing library and the audit scorecard used on every engagement.
3 senior data scientists · maintain flavor.js (3.1k★)
03
Three product designers who turn a hypothesis into a testable artifact — wireframes, prototypes, and the production build of every variant. They are senior enough to work directly inside a client's Figma library and to defend a design in front of a founder.
3 senior product designers · embedded in client Figma
04
Two experiment engineers who ship the variants, integrate the feature flags, and instrument the events. They are the reason the median cycle time from approved hypothesis to live experiment sits at nine working days.
2 senior engineers · 9-day median cycle time
412
Winning experiments shipped since 2019, drawn from 1,180 total tests run.
92%
Client retention, year over year. 71% expand scope within 90 days.
4.9/5
Average Clutch review across 63 verified client engagements.
SOC 2
Type II
Independently audited and renewed annually since 2022.
A note from the partners
We are not the largest experimentation firm. We are not trying to be. We are trying to be the one your VP of Product calls when the testing program has stalled and the board is asking why growth has flat-lined. That is a small, senior job. It needs people who have shipped the kind of experiments where the answer is sometimes no, and who can sit in a readout and say the lift is real, or the lift isn't, without flinching. Every brief we accept is a promise to do that work ourselves, not to delegate it. The size of the team is the product.
— The partners, Tyrell Lab
Recognition & Audit
The studio's reputation is built on private readouts and repeat engagements, but a few external signals are worth surfacing for anyone running a procurement or security review.
2024
Named a Leader in the G2 Spring 2024 Grid Report for Conversion Rate Optimization in the Mid-Market segment, based on verified user reviews and satisfaction scores.
2024
Featured in Animalz's 2024 editorial roundup of high-performing B2B agencies, selected on the basis of client references and shipped work rather than paid placement.
2024
Listed in Built In NYC's 2024 recognition of top-tier remote-friendly workplaces operating in the experimentation and growth analytics category.
2023
Finalist at the 2023 Pavilion Awards for Best Boutique Analytics Firm, judged on methodology, client outcomes, and team seniority.
Since 2022
Independent audit of security, availability, and confidentiality controls, renewed each year since 2022. The latest report is available under NDA on request.
2021
Backed by First Round Capital in 2021 to keep the team small, the standards high, and the engagement model senior-only. Capital without growth pressure.
Open Source
Maintained by the studio's data scientists. 3.1k GitHub stars, used by experimentation teams at several large SaaS companies for online sequential A/B analysis.
Procurement, security, or reference checks — start with a 30-minute diagnostic call.
Book a Free Experiment Diagnostic