Noteworthy posts
Posts from your recent history that carry a specific lesson. Each one is classified into a category — breakout, conversation catalyst, quiet winner, served but flat, reply magnet, format win, short & punchy, high-effort flop — and paired with Claude's plain-English takeaway grounded in your own internal benchmarks.
We’re getting flooded with accountant applications. Didn’t expect this level of response. We’re starting to shortlist candidates now. If you haven’t applied yet: this is your last window https://zipang.id/jobs/accounts-executive-dc53e189
The gap between views and engagement is consistent with the ranker distributing the post to a broader pool — possibly boosted by the earlier viral accounting post — but the audience encountering it cold found no entry point to engage. Plausibly, the low reply and like count in the first hour signaled poor resonance, causing the ranker to throttle further distribution.
UNPOPULAR OPINION : After 1.2Billion Token of GPT.5.4 Opus 4.7 < GPT5.4 xhigh Well setelah pakai claude code & codex Parallel sebulan ini, Untuk workflow gue, tahta nya dipegang sama 5.4 xhigh via Codex 🙌 Setelah kmrn tragedi Opus 4.6 di nerf gue 1 minggu kmrn jadi agak ragu sama Anthropic, downtime nya 1 minggu, lama bgt 🫣. Kalian pakai Clade/Codex buat apa btw?
6× your usual reply-to-like ratio with 9 replies on 3 likes, and the closing question ('Kalian pakai Claude/Codex buat apa btw?') is almost certainly why — a direct, low-friction question at the end of an opinion post is a proven reply driver for you. Your 907 views here is above your median of 506, so the topic had reach; the lesson is to always close AI opinion posts with a specific, easy-to-answer question.
The explicit call-to-reply at the end of the post likely lowered the friction for commenting, consistent with Meta's public emphasis on replies as the dominant Threads ranking signal. The 'unpopular opinion' framing plausibly also triggered disagreement-driven replies, which on platforms with X-like architectures carry heavier weight than passive likes.
QUICK FINDING RECRUITMENT ACCOUNTING PHASE 1 Dari screening phase 1 Banyak yang technically strong… tapi jatuh ke Tier 4 bukan karena skill. 👉 karena English level gak dicantumin Cuma ~15% yang clearly show fluent/professional English. Sisanya? CV Indo semua atau gak nunjukin levelnya. Padahal ada yang bagus banget pengalaman nya. Simple fix for next round: Tulis English level di CV, biar ga
9 replies on just 2 likes — a reply-to-like ratio 9× your usual — and notably 9 reposts, suggesting people shared this without publicly engaging. The 'Simple fix' framing gave readers something actionable to react to, which pulled replies even from a small 355-view audience. Lead with the fix or the insight earlier in the post to hook more readers before they scroll past.
The unusually high repost count relative to likes (9 reposts, 2 likes) is consistent with the post being saved or forwarded as useful information rather than liked as entertainment — plausibly the ranker received repost signals but limited reply depth, which may have capped distribution despite the strong reply-to-like ratio.
Gue baru nemu cara paling gampang jelasin bedanya OpenClaw sama Hermes. Bayangin kelas kuliah: OpenClaw = cowok paling cerewet di kelas Hermes = cowok nerd paling pendiam Mau tau detailnya? Lanjut 👇
Only 252 people saw this video, but 10 liked it — a like-per-view rate 3.2× higher than your typical posts, meaning the small audience that found it really resonated. The analogy framing ('cerewet vs nerd di kelas') made a technical comparison feel human and shareable. Consider reposting or repackaging this concept with a stronger opening hook to give it the distribution it deserved the first time.
A plausible explanation is that the post didn't generate enough early engagement velocity to trigger wider distribution — consistent with the inference that Threads (like X) surfaces posts based on first-hour signals. The high like-per-view rate suggests the content quality was there, but the initial audience pool may have been too small to create the momentum needed for the ranker to expand reach.
OVERRATED vs UNDERRATED TUI situation: - Kimi 2.6code via CLI very underrated, dipecut 60 swarm agent parallel task pake 5 terminal cuma habis 11% weekly quota WKWK - Claude Code - unusable Opus 4.6 di nerf habis2an ckckck - GPT5.4 Codex CLI worth it buat hardenning - Zai GLM Overrated
78 replies at 11× your median shows the 'OVERRATED vs UNDERRATED' format is a reliable conversation trigger for you — this is the second time it appears in your top posts. The AI tool comparison framing forces readers to pick a side or defend their stack, which is why replies nearly matched likes (78 vs 80). Treat 'OVERRATED vs UNDERRATED [topic]' as a repeatable template you can deploy monthly.
The near 1:1 reply-to-like ratio is consistent with Meta's public statements that replies are the strongest Threads ranking signal; a post that generates as many replies as likes is plausibly weighted far above a post with the same likes but few replies, since on X (a documented proxy) a reply carries ~27× the weight of a like.
Susah juga ya cari akuntan yg biasa AR/AP/GL bank reconcile (English book keeping), terbiasa komunikasi bahasa inggris, terbiasa pakai quickbooks/ xero. Remote working Gaji nya 6-10jt padahal, tapi rata-rata candidates yg masuk justru bahasa Indonesia semua akuntan nya 😭 Threads, do your magic! Nyari 3 akuntan ni 😭
This is your highest-reply post ever at 112 replies — 16× your median of 7 — and it also drove 19,561 views, your personal record. The magic was a real, relatable hiring pain shared publicly with a direct call to action ('Threads, do your magic!'). When you share a genuine business problem and ask the community to help solve it, your audience becomes participants, not just readers — do this more deliberately.
With 112 replies, this post almost certainly received a massive boost consistent with Meta's documented emphasis on replies as the strongest Threads signal. Plausibly, each reply chain extended dwell time and re-engagement, and the community-sourcing framing ('do your magic') likely lowered the barrier to commenting, compounding the reply signal early.
New remote roles dropping this week US 🇺🇸 SG 🇸🇬 JP 🇯🇵 Creative & Digital workers — be ready. English speaker only More details soon
This post hit 240 likes — 80× your median of 3 — and your personal best by raw likes. The formula was scarcity + specificity + anticipation: three countries, a clear audience ('English speaker only'), and a teaser ('More details soon') that made people save or follow. Recreate this pattern: announce something real that's coming, name the audience precisely, and leave a reason to stay tuned.
The post likely generated strong early saves and reposts (9 reposts vs your median near 0), which is consistent with Meta's public statement that conversation and re-sharing signals favor distribution. The teaser format plausibly drove return visits and follows, giving the ranker sustained engagement signals beyond the first hour.
1,082 views (2.1× your median) but only 3 likes and 3 replies — the algorithm tested this with a wider audience and they passed. The post reads as a status update for people already following the hiring story, not as a standalone hook for new readers who had no context. Next time you post a follow-up to a viral thread, write it so someone who never saw the original still has a reason to care.