Illustrative example
See the experiment branch.
CV-screening SaaS → recruiting teams · example data
Audience → channel → message → winner. The same engine works for any B2B ICP.
Audience test
Test the audience.
Test two buyer segments and let qualified meetings decide.
Audience A/B test
A
Audience A
Atlas · parent branch
Option A
Talent Acquisition Leaders
High-growth tech companies.
B
Audience B
Milo · parent branch
Option B
HR Operations Leaders
High-volume hiring companies.
Channel test
Test the channel.
Each audience splits into email vs. LinkedIn.
Channel tests — children of the audience branches
Children of Atlas
Email variant
Vega · child of Atlas
Email-first
Atlas → Email
LinkedIn variant
Lumi · child of Atlas
LinkedIn-first
Atlas → LinkedIn
Children of Milo
Email variant
Kai · child of Milo
Email-first
Milo → Email
LinkedIn variant
Rex · child of Milo
LinkedIn-first
Milo → LinkedIn
Message test
Test the message.
Each channel tests two message hypotheses.
Message tests — children of the channel branches
Results
See what wins.
Rank every path by booked, qualified meetings.
Illustrative results · Qualified meetings
1
★
HR Ops → LinkedIn → Message B
Bolt · illustrative
0
2
TA Leaders → Email → Message A
Nova · illustrative
0
3
HR Ops → Email → Message B
Ruby · illustrative
0
4
HR Ops → LinkedIn → Message A
Sprout · illustrative
0
Challenge
Then try to beat it.
The winner becomes the baseline for the next test.
★
HR Ops → LinkedIn → Message B
Champion · Bolt · 47 meetings
Current baseline
Every new variant has to beat this.
VS
C1
New CTA + timing variant
Challenger · Juno
New hypothesis
If it wins, it becomes the new baseline.