Vlix Vlix

← Digital Marketing

How to Write a Viral Tweet and Get More Followers on X

On 13 August 2026 X published the code that builds the For You feed, including the number it gives every action a reader can take on your post. You no longer have to guess what X's For You feed rewards, because X's published code says it. This guide pairs X's numbers with the only study that ever tested tweet wording properly, plus four peer reviewed papers on what people share and why.

X ranks every post in your feed on points, worked out from how likely you are to take each action. X's ranking code weighs a likely like at half a point and a likely reply at five, so it weighs a likely reply ten times as heavily as a likely like. X weighs a likely copied link, the kind a reader pastes into a chat, at 20 points, forty times as heavily as a likely like. X gives the same weight of 20 to a likely reply from somebody who follows you back. X weighs the chance that a reader reports your post at minus 234, 468 times the weight of a like, but X's notes on its ranking code say this does not mean one report cancels out 468 likes. Your second post in one reader's feed keeps only half its points. If you have 1,000 followers or fewer, X reserves a slot at position 15 of a stranger's feed for one of your original posts. So write each post to be replied to, not liked.

Illustration: a column of stacked feed cards ranked from top to bottom, each card holding a different sized pile of round counters, the fullest card sitting at the top

Understand how X scores a post, because every rule below comes straight out of it

Every score in this guide comes from the For You feed code X published on 13 August 2026.

Illustration: a set of weighing scales with one small heart on one side and a much larger stack of speech bubbles on the other

Reply to other people more than you post, because X weighs a likely reply ten times a like

What a reader doesWeight X gives the actionWeight compared with a likeCan your writing cause it
Copies your link into a chat20.040 times a likeOnly if the post is useful to one person
Replies, and follows you back20.040 times a likeYes, follow people back
Replies to your post5.010 times a likeYes, ask a real question
Quote posts your post5.010 times a likeYes, leave room to disagree
Sends your post in a direct message5.010 times a likeOnly if the post is useful to one person
Follows you from the post4.08 times a likeOnly through the profile they land on
Taps the share button2.04 times a likeOnly if the post is useful to one person
Reposts your post1.02 times a likeYes, ask people to repost it
Likes your post0.51 like, the baselineYes, agreement earns likes
Taps into your post0.40.8 times a likeYes, a cut off first line
Opens your link0.20.4 times a likeYes, links no longer cost reach
Opens your photo0.050.1 times a likeBarely worth the effort
Opens your video0.050.1 times a likeBarely worth the effort
Watches enough video to count0.050.1 times a likeBarely worth the effort
Opens your profile0.0NothingNothing you write changes this
Stares at your post0.0NothingNothing you write changes this
Blocks you, cutting you off-31.2Minus 62 times a likeAvoid, this is pure loss
Asks X for less of your posts-43.2Minus 86 times a likeAvoid, this is pure loss
Hides your posts from their feed-58.8Minus 117 times a likeAvoid, this is pure loss
Complains to X about your post-234.0Minus 468 times a likeAvoid, this is pure loss
Illustration: two figures connected by a double headed arrow, with a bright glowing speech bubble between them

Follow back everyone who replies to you, because X then weighs their reply forty times a like

Illustration: one figure handing a small card directly to a second figure, with a large empty crowd standing apart in the background

Write posts people paste into a group chat, because a copied link scores higher than anything

The three highest numbers X publishes, after the follow-back jump, are all a reader taking your post somewhere else, which you never see happen.

Illustration: a stack of identical cards where the first is full size and each one below it is visibly smaller and fainter

Post once and then stop, because X gives your second post half the points of your first

Illustration: a tall column of stacked feed cards with one small card lifted by an upward arrow into a highlighted empty slot partway down

Post originals while you are under 1,000 followers, because X saves a feed slot for small accounts

1,000 followers is the cut-off. Under it, X puts one of your original posts into position 15 of a stranger's feed.

Replies and reposts are excluded from both of the largest boosts in X's code: the follow-back reply points and the small-account feed slot. So everything that helps you most needs an original post to attach to.

A phone playing a recorded phrase in a loop while a person repeats it back, hands free

Use Repeat Recorder to rehearse the line before you film a video post

X's code gives a video view almost nothing on its own, so if you film yourself, your spoken words have to win the replies. Repeat Recorder, one of my own apps, plays a recorded phrase back to you on a loop so you can say it until it comes out smoothly, without touching the screen.

Illustration: a large red downward arrow crushing a small pile of tiny hearts beneath it

Never make a reader want to report you, because X weighs a likely report 468 times a like

X gives every reader four buttons for getting rid of you: ask for less of your kind of post, mute you, block you, or report you to X for breaking its rules.

Illustration: one rounded post card with an identical copy of it appearing beside a second blank figure, facing a row of small follower figures

Understand what a repost is, because it puts your post in front of somebody else's followers

X calls this button a repost. Twitter called it a retweet, and the research in this guide still uses that older word for the same thing.

Illustration: one figure holding out an open hand toward a circle of other figures who are each passing a small card onward

Ask readers to repost your post, because the words "please repost" were proven to work

Most claims about tweet writing compare popular tweets with unpopular ones, which mostly measures who wrote them. Researchers Chenhao Tan, Lillian Lee and Bo Pang instead found 11,000 pairs of tweets where the same person shared the same link twice using different words, so only the wording differs.

Illustration: a short stubby block of text beside a longer block shaped like a newspaper headline, with the longer one carrying an upward arrow

Write a longer post and shape it like a news headline, because both beat the short version

Illustration: a post card with several small tag labels stuck to it being pulled downward, next to a clean untagged card rising

Stop tagging accounts and stop writing the word "you", because neither one won a real test

Illustration: two cards side by side, one blank and vague, one carrying a clear numeral and a name badge, with the second one glowing

Put a real name and a real number in the post, because both work and almost nobody adds them

Illustration: a wide simple shape that fits into many different sockets, beside a narrow complicated shape that fits only one

Keep the words simple and the claim general, because plain and broad versions won the test

What you changeThe version that losesThe version that winsWhat the controlled test found
Asking for the repostNew piece on sleep.New piece on sleep. Please repost it if it helps somebody.Asking raised reposts
Length of the postInteresting study on sleep.A sleep study followed 4,000 shift workers for two years and found they lost 11 working days each.Longer version won
Tagging other accounts@alice @bob @carol you will want to read this.Shift workers lose 11 working days a year to bad sleep.Tagging lowered reposts
Writing the word "you"You are losing 11 days a year to bad sleep.Shift workers lose 11 days a year to bad sleep.The word "you" did nothing
Naming a personA researcher measured it.Chenhao Tan measured it.Names raised reposts
Using a numberLoads of tweets were tested.11,000 pairs of tweets were tested.Numbers raised reposts
Shape of the sentenceso i read this thing about sleep last night and honestly wowBad sleep costs the average shift worker 11 days a year.Headline shape won
General or specific claimThe mistake I made in 2019.A mistake most new writers make.General wording won
Reading difficultySleep deprivation precipitates measurable cognitive degradation.Losing sleep makes you think worse.Easier to read won
Hashtags on the postBad sleep costs 11 days a year. #sleep #health #productivityBad sleep costs 11 days a year.Hashtags did nothing
Illustration: a figure standing between two nearly identical cards with a large question mark above, and a coin spinning beside them

Do not trust your own taste, because people pick the better of two tweets only 61% of the time

Illustration: three simple shapes rising fast on upward arrows beside one heavy shape sinking downward

Write one high-energy feeling into the post, because awe and anger travel and sadness does not

Researchers Jonah Berger and Katherine Milkman measured every New York Times article published over three months and checked which ones got emailed on.

Illustration: a single bright word block dropped into a pool, sending a wide ring of ripples outward

Add one moral or emotional word, because each one raised the repost rate by 20%

Researcher William Brady and four colleagues read 563,312 posts about gun control, same-sex marriage and climate change, and counted the words.

A phone taking a photo, an arrow, and the same photo being pasted on a computer

Use Camera to Clipboard to get a photo off your phone and into a post

If you write your posts on a computer, the photo you want is almost always on your phone, and emailing it to yourself is how a good post turns into tomorrow's post. Camera to Clipboard, another of my own apps, puts the photo straight onto your computer's clipboard so you can paste it into the post box.

Illustration: two crowds of figures facing each other across a gap, with a single sharp arrow flying from one side into the other

Know why naming an enemy works, because one out-group word raises the odds of a share by 67%

Researchers Steve Rathje, Jay Van Bavel and Sander van der Linden read 2,730,215 posts from news accounts and members of the US Congress, and found the strongest writing effect anybody has measured.

67% higher odds of a share, for every single word naming a group the reader dislikes. It is the strongest known effect, and the worst one to build an account on.
Illustration: a fast paper plane racing far ahead of a slow heavy one, with a warning triangle trailing behind the fast one

Do not build the account on false claims, because false stories spread 70% more than true ones

Researchers Soroush Vosoughi, Deb Roy and Sinan Aral traced 126,000 news stories shared by 3 million people on Twitter over eleven years.

Illustration: a large picture frame on a set of scales weighing almost nothing against one tiny heart on the other side

Do not add a picture just to add one, because X weighs a likely photo open at a tenth of a like

Illustration: a bin swallowing a pile of small tag symbols while a chain link is lifted back up onto a card

Delete the hashtags and put the link back in the post, because X changed both rules

Illustration: a huge funnel with an enormous crowd pouring in at the top and only a handful of figures dropping out of the bottom

Do not chase one viral post for followers, because 126,000 views produced 280 followers

Tyler Hakes, a writer who tested growing an X account, published every number from two months of trying to grow an account, including the post that took off.

Illustration: a tall ordered list of bars of different lengths, sorted from longest at the top to shortest at the bottom

Every tactic in this guide ranked by the evidence behind it

Green means the evidence is a controlled experiment or X's own published code. Amber means one large dataset that watched rather than tested. Red means one account's numbers.

What to doWhere the evidence comes fromSize of the effectWhat it costs you
Follow back people who reply to youX's own ranking codeTheir reply goes from 5 points to 20A few minutes a day
Reply to other people more than you postX's own ranking codeA likely reply weighs 10 times a likeA daily habit
Write posts people paste into a group chatX's own ranking codeA likely copied link weighs 40 times a likeChanges what you write about
Post originals while under 1,000 followersX's own ranking codeA free slot at feed position 15Nothing at all
Post once instead of ten timesX's own ranking codeSecond post cut to half, rest to a quarterNothing but discipline
Never make a reader complain or hide youX's own ranking codeA likely report weighs 468 times a likeNothing at all
Ask people to repost it11,000 controlled pairs of tweetsRaised reposts, strictest test passedA few characters
Write the longer, more informative version11,000 controlled pairs of tweetsRaised repostsNothing at all
Put a real person's name in the post11,000 controlled pairs of tweetsRaised repostsNothing at all
Put a real number in the post11,000 controlled pairs of tweetsRaised repostsYou need a real fact first
Shape the sentence like a news headline11,000 controlled pairs of tweetsRaised repostsA rewrite every time
Make the post easy to read11,000 controlled pairs of tweetsRaised repostsA rewrite every time
Stop tagging other accounts in the post11,000 controlled pairs of tweetsThe only wording that lowered repostsNothing at all
Add one moral or emotional word563,312 posts, published in PNAS20% more spread per wordIt stays inside your own side
Use awe, anger or worry rather than sadnessEvery New York Times article for 3 monthsRaised sharingDepends on your subject
Name a group your readers dislike2,730,215 posts, published in PNAS67% higher odds of a share per wordFollowers who came for a fight
Drop the hashtags11,000 controlled pairs of tweetsNo effect either wayNothing at all
Add a picture to every postX's own ranking codeA likely photo open weighs a tenth of a likeTime you could have spent writing
Chase one viral post for followersOne account's published numbers280 followers from 126,000 viewsWeeks of chasing
Illustration: a figure on a hilltop at the end of a winding path, holding a green check mark above their head

Conclusion: how to write a viral tweet and get followers, in one glance

Write posts to be replied to, not liked. These six rules are ordered by the size of the measured effect.