Turn hypotheses into conversations you can learn from.
- 01
Hypothesis
Segment, trigger and claim fixed before writing.
- 02
Sequence
Multichannel steps with one variable isolated.
- 03
Reply
Positive, objection and silence read separately.
- 04
Learning
ICP, data scoring and copy updated.
Output feeds the next cycle
Illustrative example — not client work.
How a campaign is actually built
Campaigns are launched to fill the calendar, not to prove anything.
- One explicit hypothesis per campaign
- Segment, persona and trigger fixed before writing
- The result that would change the plan, stated up front
Campaigns that produce a decision either way.
The hypothesis becomes copy and sequence structure.
Campaigns are launched to fill the calendar, not to prove anything.
- One explicit hypothesis per campaign
- Segment, persona and trigger fixed before writing
- The result that would change the plan, stated up front
Campaigns that produce a decision either way.
The hypothesis becomes copy and sequence structure.
Messages describe the product and read like every other email in the inbox.
- Problem-first copy per segment, short and specific
- Multichannel steps across email and LinkedIn
- Variant structure that isolates one variable at a time
Sequences whose results can be attributed to something.
Live sending is governed by infrastructure limits.
Nobody owns the week-to-week operation, so campaigns drift.
- Scheduled launches within deliverability caps
- Daily monitoring of bounces, placement and replies
- Fast kill or scale decisions on weak variants
A motion that runs whether or not the founder is selling this week.
Replies move into qualification.
Results are reported as open rates rather than market feedback.
- Weekly read of what the market confirmed or denied
- Message and segment adjustments with a written rationale
- Findings fed back into ICP and data scoring
A compounding view of which market actually responds.
Feeds the next strategy cycle.
Frequently asked
Enough to read a result per segment, not so many that nothing reaches significance. Volume is set by the decision you want to make.
Where it improves relevance against a real signal. Generated flattery with no research behind it performs worse than a plain, specific message.
Find the constraint before you scale it.
A five-question self-assessment that suggests a starting point and the sequence that follows it.