A free tweet analyzer that scores your tweet against 11,118 real X tweets, shows it in X's timeline, and reads any tweet's numbers from its link.
5.0(1 rating)Write a tweet or paste a link to one. You see it in X's timeline, scored by a model trained on 11,118 real X tweets, and checked part by part.
Your tweet
For a thread, put a line with --- between posts.
How it looks on X
Post: photo, picture, name and badge
The score
What the tweets say
On 280 accounts it never saw, the tweet with the higher score got more likes 58% of the time, in 30,483 pairs with a clear winner. A coin flip gets half, so use the score to pick between versions you like. Where the data comes from · It reads English tweets.
Make it better
Count its charactersWhat you write stays in this browser, and the score runs on your device. A link is read on my server, and only the tweet and its numbers come back.
This free tweet analyzer scores your tweet with a model trained on 11,118 real X tweets from 298 accounts.
Write a tweet or paste a link to one. You see it in X's own timeline, on a real tweet or on your own account, with the parts that helped or hurt in real tweets. A link also gets its real numbers against the account's typical tweet.
Here's what a tweet analyzer is, how the score works, what 11,118 X tweets show, and how to write tweets that beat your own usual.
What is a tweet analyzer?
A tweet analyzer checks a tweet before you post it, or after, and shows what to change.
Most score against rules someone wrote, or ask an AI to guess. This one scores against what won: tweets from the same account, posted within 30 days of each other, where one got at least twice the likes of the other.
How the score works
The analyzer measures 25 parts of your tweet: its first line, its length, numbers, lists, "I" and "you", links, hashtags, mentions, pictures and more.
A model weighs each part by how often it won those matchups. Your score says where your tweet lands among all 11,118: 70 means it beats 70% of them.
How accurate it is
I tested it 5 times. Each time the model learned from 4 in 5 of the 280 accounts and was graded on the fifth, which it never saw, so every account was graded once.
Across those tests, the tweet it scored higher got more likes 58% of the time, in 30,483 pairs where one tweet got at least twice the other's likes. A coin flip gets 50%.
That's a hint, not a promise. The topic, the timing and the picture itself aren't in the score, and they matter more than the wording.
What 11,118 X tweets show
In pairs of tweets from the same account, where one got at least twice the likes, this is how often the tweet with each part won:
- Won
- 59%
- Pairs
- 8,258
- Won
- 57%
- Pairs
- 621
- Won
- 57%
- Pairs
- 4,341
- Won
- 56%
- Pairs
- 5,505
- Won
- 55%
- Pairs
- 12,373
- Won
- 55%
- Pairs
- 8,377
- Won
- 54%
- Pairs
- 2,064
- Won
- 53%
- Pairs
- 12,367
- Won
- 53%
- Pairs
- 10,945
- Won
- 47%
- Pairs
- 5,456
- Won
- 46%
- Pairs
- 10,029
- Won
- 45%
- Pairs
- 1,285
- Won
- 45%
- Pairs
- 2,765
- Won
- 44%
- Pairs
- 8,489
- Won
- 39%
- Pairs
- 7,187
- Won
- 39%
- Pairs
- 4,575
- Won
- 35%
- Pairs
- 6,208
The biggest winners: a long post (59%), a thread mark (57%) and a list (57%). The biggest losers: a link (35%), a question in the first line (39%) and a quote of another post (39%).
Where X cuts a long post
A long post shows its first 276 to 280 characters in the timeline, cut at the end of a word, then Show more. I read that from X's own data for 5 long posts.
So the first 280 characters have to earn the click. Put the point there, not the setup.
- Limit
- 280 characters
- Source
- X's counting rules
- Limit
- 25,000 characters
- Source
- X Help: X Premium
- Limit
- About 280 characters, cut at the end of a word
- Source
- X's own data for 5 long posts, read 2026-09-27
- Limit
- 23 characters, however long
- Source
- X's counting rules
- Limit
- 2 characters
- Source
- X's counting rules
The analyzer counts your post the way X does, and draws it in X's timeline, cut where X cuts it. For the count alone, the X character counter splits a long draft into a thread for you.
Read any tweet from its link
Paste a link to any public X tweet, yours or anyone's, and the analyzer reads it for you: the full text, the pictures, the likes, reposts, replies, bookmarks and views.
Then it reads the account's latest tweets and compares. "4.1× this account's typical tweet" means the tweet got 4.1 times the median likes of the account's recent tweets.
That's the honest version of a virality score: not a guess about the wording, but how the tweet actually did against the account's own normal.
How to write tweets that beat your usual
Here's what won most often in the data:
- Go past 280, if you have X Premium. The first 280 characters show, and the rest rewards the people who tap. Tweets with it won 59% of 8,258 pairs.
- Break it into lines. A list of short lines reads faster than a paragraph. Tweets with it won 57% of 4,341 pairs.
- Add an emoji or 2. Used as a marker, not decoration. Tweets with it won 56% of 5,505 pairs.
- Tell it as your own story. "I" and "my" make it yours, not advice from nowhere. Tweets with it won 55% of 12,373 pairs.
- Point at the answer. "Here's" and "this" promise something is coming. Tweets with it won 53% of 10,945 pairs.
And what lost:
- Put the link in a reply. Tweets with a link won only 35% of 6,208 pairs.
- Answer instead of asking in the first line. Tweets that open with a question won only 39% of 4,575 pairs.
- Post your own take instead of quoting. Tweets that quote another post won only 39% of 7,187 pairs.
- Make a statement. Tweets with a question won only 44% of 8,489 pairs.
Then post, and let your own numbers decide. Paste the link back in after a few days to see how it did against your usual.
The Art and Business of Online Writing
by Nicolas Cole
The Art and Business of Online Writing by Nicolas Cole reveals how to write compelling content, grow an audience.
I may earn a commission if you make a purchase, at no additional cost to you.
Where the data comes from
I read the latest public tweets of 298 X accounts: creators from my creator directory, the creators in my research feed, and well-known creators in business, writing, AI, health and money. My own accounts are left out.
Replies, reposts and tweets under 3 days old are left out too, because their numbers are still growing. That left 11,118 tweets.
The model learned from pairs of tweets by the same account within 30 days, so the audience was the same and the wording was the difference. Each account was graded by a version of the model that never saw it. The tweets themselves stay private: only the model's weights and counts are published.
Ready to post it?
Or post it from your AI with my open source MCP servers and CLIs
Tweet Analyzer FAQs
Questions about the tweet analyzer? Here's what to know.
A tweet analyzer checks a tweet and shows what helps or hurts it.
This one scores against 11,118 real X tweets, not rules someone wrote.
It measures 25 parts of the tweet, like the first line, numbers, lists, links and pictures.
A model trained on pairs of tweets from the same account weighs each part.
On 280 accounts it never saw, the tweet it scored higher got more likes 58% of the time, in 30,483 clear pairs.
A coin flip gets 50%, so use it to pick between versions you like.
In the data, tweets did better with a long post (59%), a thread mark (57%) and a list (57%).
They did worse with a link (35%), a question in the first line (39%) and a quote of another post (39%).
In the data, a tweet with a link won only 35% of 6,208 pairs against a tweet from the same account without one.
Put the link in a reply instead.
X shows the first 276 to 280 characters of a long post in the timeline, cut at the end of a word, then Show more.
The analyzer draws your post cut the same way.
In the data, a post longer than 280 characters won 59% of 8,258 pairs against a shorter post from the same account.
Long posts need X Premium.
Yes, paste a link to any public X tweet.
The tweet analyzer reads its numbers and compares them with the account's recent tweets.
Yes, paste your X link or username under the card and it pulls your photo and name.
You can also set them by hand.
Yes, the tweet analyzer is free, with no sign-up.
The score runs on your device.
What you write stays in your browser, so it's there when you come back.
A link is read on my server, and only the tweet and its numbers come back.
Navid Moazzez
AI business strategist & AI OS builder
Navid Moazzez helps creators and founders master AI and build their own AI Operating System (AI OS) to automate their business and life.
Navid.me is reader-supported. When you buy through links on this site, I may earn an affiliate commission. Learn more.
More free tools
LinkedIn
Mastodon
Instagram
TikTok
Pinterest
YouTube
Amazon KDP
Apple Books
Amazon
Substack
Spotify
Apple Podcasts
Free AI newsletterThe most actionable AI newsletter for founders
Every week, get proven AI strategies, curated tools, and step-by-step systems to grow your audience, create better content, and build a profitable creator business.
No fluff, no filler, no BS. Just five minutes each week that might level up your online business and life.
P.S. Sign up now to get free access to my ultimate AI tools guide for creators.
