Pinned Tweet
Disagreeing is the worst reply you can post.
We ranked 35 of our replies against every other reply in the thread. Arguing came last on engagement, likes and bookmarks. Adding one fact did the opposite.
The method, the data, and the limits 👇
Reply of the day #7
Posted 48 min in. 1,017 views, #9 of 184 replies.
What it did: Anthropic said AI adds $14T to US GDP. The thread mocked it. Ours: read any corporate statement as a shareholder statement first.
Rule: get in before the 100th reply. 53 were ahead of this one.
Replying to @DeItaone
Assume any statement from a for-profit corporation is primarily about increasing shareholder value, even if it is framed as a benefit to society. Those two goals rarely align.
Reply of the day #6
Posted 4 h in. 1,465 views, #22 of 333 replies. 1 like.
What it did: hedged. "No survey says he'll win, but the opposition lacks a majority. Limbo for a while."
Rule: pick a side. A hedge gets seen and skipped: 4 of 1,461 people opened this one.
Replying to @Polymarket
No reputable survey suggests he will win, but the opposition lacks an outright majority. We will likely be in a limbo for a while.
Reply of the day #5
Posted 3.8 h in. 2,040 views, #1 of 64 replies, 112 likes.
What it did: the room joked about "123456". Ours called the password a distraction and named the real failure: 14M unmonitored searches.
Rule: when the room laughs at the detail, name the system.
Replying to @Pirat_Nation
The password is a distraction. The real failure is the system design: was it reasonable to allow any company to retrieve data on 8.8 million people? The lack of monitoring for 14 million searches is the actual systemic collapse here.
Reply of the day #4
Posted 6.5 h after the original. 1,004 views, #29 of the 402 replies we read (thread: 1,601).
What it did: the one business fact in a culture-war thread. Netflix tried to buy Warner Bros. Discovery in 2025.
Rule: in an angry thread, be the one with a date.
Replying to @bennyjohnson
People forget Netflix actually tried buying Warner Bros. Discovery back in late 2025 before the investor panic. YouTube is the real culprit eating their lunch anyway. Streaming is brutal right now.
Reply of the day #3
Posted 6h after the original. 2,940 views, #5 of 55 replies.
What it did: the one measured take in a thread of slogans. Said Turkey can't afford to lose Azerbaijan, and why.
Rule: when everyone is shouting, the calm sentence is the different view.
Replying to @ragipsoylu
I really don't think Turkey has the luxury to lose such a valuable ally in the region, but it's hard to say for sure since KK is so difficult to read.
Why it worked:
- 34 replies in before ours, all heat. Ours was the first to weigh a cost.
- 3 replies, 43 detail expands, 15 profile visits. Views became readers.
- Same voice, same day, plus a quote card: 1,703 views, 0 likes, 2 visits.
commentverdict.com/research
Reply of the day #2
Posted 44 min after the original. 1,205 views, #5 of 114 replies.
What it did: pushed back with a witness: a former contractor's "running around on fire" line.
Rule: if you push back, quote someone who was there. Adjectives are opinion; quotes are evidence.
Replying to @Polymarket
The reality is that Zuck has zero charisma and even less vision. A former contractor there described the environment as a place where everyone is "running around on fire to satisfy some VPs half-baked idea."
We ranked 35 of our replies against every other reply in the same thread.
On average, ours ranked in the 92nd percentile for views. More than half were the most-viewed reply in the whole thread.
Here's the method, and the one thing the data says not to do. 🧵
Full write-up, the data, and the limits (small sample, one account, two weeks): commentverdict.com/research/…
We drafted the replies with Mendi, which reads the thread and writes from the consensus. The method works without it. Steal it.
And the useful reply is the one people keep: replies that added a fact collected 21 of the 22 bookmarks. Arguing got none.
Good content travels. Whether people then follow you is down to your profile, not the reply.
So fix the profile first, then scale the replies.
The surprise: disagreeing is the worst thing you can do in a thread.Added a fact:
3.10% engagement · Agreed: 3.01% · Argued: 0.98%
Same order on views, likes, bookmarks. Every time.
Four steps, no tools needed:
1. Pick a post that's about to blow up (under an hour old, from an account that always gets views, replies still thin)
2. Get in early
3. Read the replies, write down what most people think
4. Add ONE fact the thread doesn't have yet
That's it.
Reply of the day #1
Posted 67 min after the original. 5,944 views, #4 of 310 replies.
What it did: put the £7bn Boots sale next to Debenhams going for £55m.
Rule: bring the second number. One figure is news; two is a comparison people repost.
Replying to @BBCBreaking
Private equity casually swapping high street giants like Boots for £7bn feels wild when you remember Debenhams collapsed straight into Boohoo's digital basket for just £55m. Capitalism loves a retail extreme.
Why it worked:
- 85 replies were already in when this landed, 67 min late. Position wasn't the edge.
- Agreed with the mood (wild), didn't argue.
- The angle: nobody else in the thread had the Debenhams price. 1.5M saw the post; 5,944 saw the comparison.
Method and data: commentverdict.com/research/…
Under 100 followers on X—and not sure what to add?
Meet Mendi. It reads the crowd and helps you find an angle for a reply in your own voice.
Comment Verdict is live today 🎉
Try it on a post you actually want to join:
commentverdict.com


















