Writing
What we argue about
in Karlín.
12 pieces on growth engineering, positioning, product scope and what actually transferred between three exits. Everything here is material we work through in the room over a weekend.
01
Growth Engineering
Building the machine that acquires customers, rather than doing the acquiring by hand.
How to tell which growth idea actually worked
A working instrumentation setup for a small team: twenty events, one identity graph, and the three dashboards that answer the questions you will actually be asked.
Your reply rate is a data quality metric
Deliverability, list building and the specific checks that move a cold campaign from 1% replies to 8%, written for engineers who would rather fix a pipeline than write copy.
What a two-person team should build and what it should rent
A concrete line between the systems worth wiring together yourself and the ones that will eat a quarter, with the maintenance arithmetic that decides each case.
02
Positioning
Deciding what you are, for whom, and against what — before you write a line of copy.
Your homepage is vague because you won't name the alternative
How to find the thing your buyer actually does instead of buying you, and how to put it on the page without sounding like a comparison chart.
Your price is read before your copy, and it makes a claim
A price tag tells a buyer who the product is for and what happens if it fails, long before they read a word. Here is how to set one deliberately.
How to run ten customer calls that actually change what you build
A recruiting rule, four questions that produce facts instead of opinions, and the coding step that turns ten transcripts into one roadmap decision.
03
Product
Shipping the smallest thing that proves the thesis, then the next smallest thing.
How to cut a two-week slice that still answers something
A method for shrinking scope without shrinking the question, so that a fortnight of build produces a decision you can defend rather than a demo you can show.
Onboarding is the product and the first eight minutes decide
What actually happens in a new user's first session, why the drop-off is almost never where founders think, and how to instrument and fix it in a week.
The maintenance tax nobody prices in when killing features
How to find the features costing you more than they return, what the real carrying cost of a feature is, and a deprecation sequence that does not burn trust.
04
Founder Notes
Three exits, several wrong turns, and what actually transferred between them.
What I got wrong hiring my first engineer, twice
Two first engineering hires, two different failures, and the specific interview and onboarding changes that stopped me repeating either of them.
The month revenue was real and the business was not
How £61,000 of genuine invoiced revenue hid a business that was shrinking, and the four numbers I now check before I let myself believe a good month.
Running a company from Prague, costed honestly
The real arithmetic of operating from Prague: what the timezone actually buys you, where the hiring pool runs thin, and the three things that are genuinely harder.
The list
One letter a month, on the same subjects as the writing.
What I am seeing across the weekends: what is working in growth engineering, what stopped working, and the numbers behind both. No sequence, no upsell ladder, and one click to leave.