This week was one of those rare ones where the numbers and the narrative actually lined up. I published 8 posts across AI, leadership, and process design, and the data from GA4 tells me something interesting: the posts that took the least time to write are the ones people actually read.
Let me break down the week.
What I Built and Learned
Monday kicked off with a GROW coaching experiment in my 1:1s — a framework I had been skeptical about for months. I wrote it up, hit publish, and moved on. By Wednesday, the analytics showed people were spending over 3 minutes on that post. That never happens with my technical content. Turns out, the leadership and process pieces resonate harder than the deep dives into gRPC vs REST or homelab cost optimization.
On the AI side, I shipped two posts I am genuinely proud of: “One Markdown File Made My AI Agent 23 Points Smarter” and “Homelab AI Agent Costs Down 60% with Ollama Quantized Models.” Both came from real experiments in my home lab — no theory, no vendor benchmarks. The markdown file post is basically a pattern I use daily now: a single system prompt file that acts as the agent’s persona definition. The homelab post documents how I swapped full-precision models for quantized ones and watched my electricity bill drop while inference speed stayed usable.
I also wrote about responsible AI from a builder’s perspective — not the ethics committee version, but the practical trade-offs you make when you decide which guardrails to ship and which to skip.
The Numbers
8 posts published across 4 categories. GA4 tracked 243 page views for the week, with Monday and Tuesday being the strongest days (47 and 49 views respectively). Traffic was mostly direct and organic social — organic search is still quiet, which tracks with the GSC data showing only 8 impressions for the period (Google Search Console has a 2-3 day lag, so last week’s posts are still indexing).
Category breakdown: AI and General each pulled about 65 views, with IoT and CyberSecurity trailing. That matches the content mix — I wrote more broadly this week rather than going deep into a single niche.
Highlight: The GROW Coaching Post
The best-performing post by engagement was “I Tried GROW Coaching in My 1:1s. It Cut Them in Half.” Six sessions, but an average session duration of 210 seconds and a 60% bounce rate. That means people who clicked through actually read the whole thing. The takeaway for me: practical, personal stories about process changes outperform technical explainers every time. I spent 45 minutes writing that post. I spent 3 hours on the gRPC vs REST comparison. The 45-minute post got 3x the engagement.
Lesson learned: lead with the story, not the architecture diagram.
What is Next
Next week I am doubling down on the format that worked. More short, practical posts about what I actually built and what broke. I have a few experiments queued up: running a local RAG pipeline with quantized embeddings, a comparison of MCP server implementations for home automation, and a follow-up on the GROW framework applied to code reviews instead of 1:1s.
If you read one thing from this week, make it the GROW coaching post. It might save you an hour of meeting time per week.
Related reading:
Discover more from Susiloharjo
Subscribe to get the latest posts sent to your email.