I started Builder's Compass in late 2023 because I was tired of giving the same architectural advice in five different DMs every week.

I would have a long-form conversation with a senior engineer about whether to extract a microservice. The next week a different engineer would ask the same question with different specifics. The week after that, a third. The advice was the same; the names of the variables changed. I could not scale the conversation, and the conversation was the most useful thing I was doing professionally.

So I wrote the conversation down and emailed it to twelve people. Then to forty. Then to four hundred. Now to about three thousand eight hundred. Once a week, every week, for the better part of two years and a quarter.

I am writing this on a Tuesday in mid-January, four days after Studio shipped, eighteen weeks into building DuranteOS. The newsletter and the product are entangled in ways I did not expect when I started either one. This post is the retrospective I have been promising my list. It is also the playbook for any technical founder who is wondering whether writing publicly is worth the effort. Spoiler: it is, but not for the reasons you would guess from the outside.

The numbers, honestly

I will start with the numbers because every newsletter retrospective opens with vague vibes about "engagement" and I want to be specific.

A few things to notice. First, the time-per-issue went up, not down. Faster writing did not happen. Better writing did. Second, the open rate is high (industry benchmarks for technical newsletters are 25-35%), and the click-through rate is anomalously high — both signals that the list is engaged rather than inflated. Third, sponsorship demand is real but not the primary value of the list.

Fourth, and this is the part that took me the longest to internalize: the growth has been slow on purpose. I have crossed two thousand subscribers, three thousand subscribers, three thousand five hundred subscribers, all in the same year. The compounding is invisible exactly when you are inside it. From the outside it looks like steady drip; from the inside each crossing felt like nothing was happening for months until something was.

What the list is actually for

The single biggest mistake I see technical founders make about newsletters is treating the list as a marketing channel for whatever they are selling. It is not. The list is a thinking surface.

When I sit down to write Builder's Compass, the audience forces me to take an architectural intuition I have been carrying around for two months and turn it into something defensible in 1,800 words. The discipline of writing it down — knowing nearly four thousand sharp readers will scrutinize it — is what converts vague intuition into stated thesis. The newsletter is the unit test for my own ideas.

The marketing benefit is real but secondary. The thinking benefit is the actual product. If the list went to zero subscribers tomorrow, I would still write the issues. I have empirically proven this — I wrote four issues to twelve people before the growth started, and the act of writing them was already valuable.

The five patterns that worked

Across roughly one hundred and fifteen issues I have shipped, five patterns have consistently produced the highest-engagement issues. None of them are exotic. All of them are difficult.

The four patterns that did not work

I want to name these because most newsletter advice describes only what worked.

The one piece of advice I keep giving away

When founders ask me about starting a newsletter, the advice that compounds the most is also the cheapest: publish for one year before you measure anything.

The first year is when the writing voice forms. It is when you discover which topics you can actually sustain. It is when the audience you accidentally attract teaches you who you should have been writing for. Measuring growth in months 1-12 will tell you almost nothing. Measuring quality of replies will tell you everything — but only because the replies are how you find your actual audience.

I had 400 subscribers at the end of year one. I have about 3,800 at the end of year two and a quarter. The growth was slow on purpose. The slow growth gave me time to figure out the voice without optimizing for vanity metrics. Founders who optimize for vanity metrics in months 1-12 build lists they cannot sustain in years 2-5; founders who optimize for reply quality in months 1-12 build lists that compound past month 24 without noticing.

What this implies for DOS

The connection back to my product work has been visible to me for about three months and visible from the outside for about three weeks. Builder's Compass is the single most efficient distribution channel for DuranteOS that I have access to. It does not work as a sales channel — direct CTAs to "buy DOS" would betray the trust the list has built — but it works as a credibility channel. Every issue is a small evidence of the architectural taste DOS embodies. Readers who are starting to ask about DOS this quarter tell me the same thing: "I subscribed for the writing; the writing is making me believe the product is real."

That is the asymmetric value of writing publicly. It does not sell the product directly. It makes the product unable to be unbelievable.

The dispatch column you are reading on this blog is the second movement of the same instrument. The newsletter is for sustained architectural argument; the dispatch is for weekly news synthesis. They share the same operator and the same voice but they serve different jobs. Splitting them was deliberate — a single channel trying to do both jobs would do neither well.

If you are a technical founder considering whether to start writing publicly, the answer is yes, but not for the reasons your growth-marketer friend will give you. Write because you need a thinking surface. Write because the discipline of weekly publishing forces ideas into shape. Write because the readers you accidentally attract will teach you who you actually are professionally.

The list will grow as a side effect. The thinking will grow as the actual product. The compound returns to both will arrive in year three, on a Tuesday morning, when you sit down to write the issue and realize you have something real to say that you would not have had if you had not been writing every week for two years.

I am writing the next issue tomorrow.

You can subscribe here if you want to read it.