TechThatMattRs: Week Two of Building a Content Engine in Public

I got laid off in December.

Startups do startup things. Not a surprise. And if you are reading this from a similar spot right now, you already know what the market feels like. The queue is long, the responses are slow, and the roles that actually fit are fewer than the job boards make them look.

So you do what you are supposed to do. Update the resume. Work the network. Log the applications. I am well into triple digits across a large pool of companies spanning multiple verticals. That work runs every day, because unemployment is not a sabbatical.

But a resume tells a hiring manager what you did. A content campaign shows them how you think. Most PMMs in a job search do not build one in parallel, not because they cannot, but because it takes a different kind of energy to ship work when nobody asked for it, on a timeline you set yourself, with no team and no brief. That is the extra step. And in a market this crowded, it is the differentiator.

So I built one.

This is the first status update: what shipped, what slipped, what the numbers say, and what I am changing next. You are not just reading this. You are the stakeholder I am accountable to. If something stops making sense, say so.

Here is where things stand.

What I Am Building and Why

TechThatMattRs is for people who have to operate the thing, explain the thing, or both. Security practitioners, identity engineers, SecOps leads, DevOps teams, ITPros, platform owners, PMMs, sales engineers, and GTM teams are trying to tell a coherent story without turning it into vendor noise. That is the audience. The topics follow the real operational problems, not the hype cycle.

Right now, my focus is on agent identity and governance, because the gap between what organizations are doing and what they need to be doing is still wide. Agents are identities, and most organizations are not treating them with the same rigor they already apply to human access, service accounts, or app registrations yet. So, that is where I am starting.

But it will not be the only place I go.

The Setup

Asana is the backbone. Every deliverable, every publish date, every SEO task, every featured image, every UTM string lives as a tracked task. The workflow is intentional: draft, optimize, tag, image, schedule, post, distribute. Each piece of content goes through the same checklist, whether it is a technical post or a LinkedIn social.

WordPress is the delivery engine. One canonical home for everything long-form. I tried resurrecting an old LiveJournal blog first, hit the platform limits fast, and moved everything over. Not a hard call. WordPress gives me a clean URL structure, meta descriptions on every post, and UTM tracking on every outbound link, so I can see what is actually driving clicks.

LinkedIn is the social layer. The post is the question. The blog is the answer. That framing matters because early data already showed what happens when a post gives away too much: the reader gets what they came for and stays on LinkedIn. The click only happens when there is still something left to go get.

Three content tracks run in parallel: technical practitioner posts, long-form argument pieces, and meta PMM posts like this one.

43 deliverables. 6 content types. 14 weeks. One person.

The math is 2.7 tasks per week, and everything has shipped on time. Where I am behind is board hygiene. Task closures in Asana are lagging actual publish dates, which is the predictable friction of being both the person doing the work and the person updating the system.

The Numbers

Two weeks in. Here is what is actually happening.

The blog is early, and the raw numbers look it. 36 views, 28 unique visitors. Without GA4, I cannot tell you where the traffic is coming from. WordPress’s free tier does not break that out, so attribution is limited to patterns and educated guesses right now. I own that gap.

LinkedIn is where the signal is cleaner and, honestly, where it matters most at this stage.

965 impressions in the last seven days, up 104% week over week. But the raw number is not the interesting part. The interesting part is who is in it. Senior practitioners, directors, CXOs, and enterprise companies. Twelve percent of the audience works at Microsoft. Two weeks in, no paid distribution, and the right people are already finding it. That is the number that actually matters.

This number that is currently worth paying attention to is “sends”.

Shadow Agents got 4 private sends and zero public comments. Think of it like a conference hallway. A reaction is someone nodding as they walk past. A send is someone pulling a colleague aside and saying you need to read this. It costs social capital, it happens in private, and it means the content landed with enough weight that someone wanted to put their name behind it. Four sends at this audience size, this early, is a stronger signal than four times as many reactions would be.

The thing I am watching is whether that ratio holds as reach grows. If it does, there is a word-of-mouth engine building underneath the surface numbers. If it sends dry up while impressions climb, the content is finding a bigger but less relevant audience. That is the metric I will report on every week.

Now the gaps, because there are real ones.

The conversion problem is not fixed yet. Impressions are growing. Blog clicks are not keeping pace. The LinkedIn posts are still giving away too much of the answer on the platform. The reader gets what they came for and stays put. I know what the fix is: write the post so it creates the question, not answers it. I am still calibrating how to do that consistently without it feeling like a tease.

No GA4 means no clean attribution story yet. Budget constraint, not an oversight, but it is still a gap, and it limits what I can prove about organic reach right now.

And outside of Field Notes, the content has not sparked much public conversation yet. The sends tell me people are talking about it privately. Getting that into the comments is a later problem, but worth naming.

The AI in the Room

There is a weird tension around AI in the PMM world right now. Use it too visibly, and people assume you cannot do the work yourself. Do not use it, and you are probably falling behind. Most people just quietly use it and say nothing.

That said, here is what I am doing for this work.

Most of the early foundation work for this campaign happened in ChatGPT. Long sessions, deep context, and genuinely good output once a session found its footing. The problem is performance. As sessions grow and context builds up, response times degrade badly. What should take seconds starts taking minutes. At a certain point, the session becomes too slow to be worth continuing, and you are starting over anyway.

I shifted to Claude for this phase. My honest assessment: mid. It holds structure across a long session, catches when a line is recycling from an earlier post (when called out), and can move fast once the direction is crystal clear.

The thinking was always going to be mine. The AI reflects your point of view back at you. The stronger the point of view going in, the better the output coming out. AI collapses the distance between idea and draft. It does not collapse the distance between draft and good. That gap is still entirely human. And that gap is also the job.

The image up top is also AI-generated (obv). I use a prompt generator to build a detailed scene brief, run about a dozen iterations, and pick the one that is closest to right. That one took maybe twenty minutes and four rounds of prompting. The guy in the image is not me exactly, but it is not far off either. Dual monitors. Energy drink. Gummy worms. Tired but still at the desk. Classic Matt.

What is shipping next week

This post drops on Monday. Tuesday, Approved Tool, Expanding Agent publishes. That one has the most complete governance argument I have shipped in this campaign so far, and the LinkedIn post for it will be the first real test of whether I have fixed the conversion problem.

Later in the week is TBD. There is a potential reactive post taking shape around the Stryker attack and the identity angles in it. If the reporting holds up and I can write the honest version before the news cycle moves on, that goes out. If not, it is a back-catalog spotlight. Either way, something ships.

I also need to assign dates to the distribution tasks. That one has been sitting on the board without a due date since week one. That is on me, and the clock is running on it being a sequencing choice versus just being late.

See you, space cowboy.