Growth audit
Funnel analysis, cohort retention curves and a prioritized backlog of 40 ranked experiments.
Growth teams since 2018
Run more experiments than you can staff for
We act as your growth team: a weekly test cadence across acquisition, onboarding and retention, with a written record of what worked and what did not.
7 yrsrunning growth programs
3,100experiments run
27%of tests produce a win
18active engagements
Where we work
Acquisition, activation, monetization, retention
We start where the leak is largest, which is rarely the top of the funnel.
Embedded growth team
A strategist, an analyst and a designer running a weekly test cycle against your backlog.
Onboarding and activation sprint
Eight weeks focused only on getting new users to their first real outcome faster.
Pricing and packaging study
Willingness to pay research and plan structure modeling, tested with live cohorts where possible.
Analytics instrumentation
Event tracking and cohort reporting so experiments can be read without argument.
Our method
We publish the losing tests too
Compound was built by growth leads from three subscription companies who were tired of agencies reporting only their wins. Every client gets a running experiment log with the hypothesis, the result and the call we made. Roughly a quarter of tests win, which is normal, and pretending otherwise makes teams stop testing.
•Weekly test cadence with results reviewed every Friday
•Full experiment log shared with your team, wins and losses
•Statistical thresholds agreed before a test starts, not after
Reviews
Client outcomes
“Trial to paid went from 14 percent to 23 percent over two quarters, almost entirely from onboarding changes rather than more traffic.”
Hana Y., VP Growth, SaaS
“They talked us out of a pricing change we were sure about. The willingness to pay study showed we would have lost our best segment.”
Doug E., Founder, subscription box
FAQ
About experiment led growth
How many experiments can you actually run?
Typically two to four meaningful tests per week depending on your traffic. Low traffic products need bigger swings and longer read windows, and we will tell you upfront if your volume cannot support standard significance testing.
What if most tests fail?
Most will. A healthy program wins on roughly one in four tests, and the wins compound while the losses cost only the time to run them. If we ever report a run of tests where nearly everything won, something is wrong with the measurement.
Do you need engineering time from our team?
Some. We can ship many tests through a testing tool and existing settings, but the highest impact experiments usually touch the product. We size engineering effort in the backlog so you can prioritize against your roadmap.
How long is the minimum engagement?
Six months. A shorter run does not give enough completed test cycles to see compounding, and you would be paying for setup without the payoff.
Contact
Find the biggest leak first
Share read access to your analytics and billing data. The audit will show you where users are actually falling out.
Phone(799) 555-0199
Address350 South Main Street, Suite 1120, Salt Lake City, Utah
HoursMon to Fri 8am to 5pm MT
AreaRemote, subscription businesses across North America