Social Media Hooks: 20 First Lines From My Own Threads Posts, Ranked by Reach!
Social media hooks decide whether a post gets read, so I stopped guessing and sorted mine.
I pulled my last 600 Threads posts, 666,169 views in total, and ranked them. Then I cut every post down to its first line.
Every one of the top 25 was a thread. Not one was a question. Sixteen of the 25 started with a number.
Below are 20 of the first lines that earned the most reach, the five patterns they share, and a prompt that runs the same check on your own account in ten minutes.
How I Ranked These Social Media Hooks
The window runs from 10 August to 19 October 2026, on my main account, dkspeaks. The data comes from my own account on Threads, through a connection that returns every published post with its views, replies and reposts.
I split each post on its first line break. That first line is the hook. It is all a stranger sees before deciding to open the post or scroll.
Then I grouped the hooks by feature. Does it start with a number? Is it a question? Does it start with “I”? Is it a thread or a single post?
For each group I counted three things. The share of my posts, the share of my total views, and how that group’s average compares with the average post.
That last number matters most. A group at 1.0x is average. A group at 3x is earning three times what a typical post earns.
I will give you the groups as ratios and ranks, because a ratio travels. Your account is not my account, but a pattern that wins at 3x on mine is worth testing on yours.
The 20 First Lines, Ranked by Reach
Here are 20 first lines from the top 25, in order. I left out the top three, which share one opener I keep to myself, and one repeat of a line that already appears.
Views are Threads views at the time I pulled the data, rounded to the nearest whole number.
RANK VIEWS FIRST LINE
1 29,226 5 THINGS I WILL NEVER ASK CLAUDE TO DO AGAIN
2 13,193 5 THINGS I REMOVE BEFORE I PASTE ANYTHING INTO CLAUDE
3 9,992 5 THINGS I NEVER PASTE INTO CLAUDE, AND WHAT I PASTE INSTEAD.
4 9,347 I BANNED 5 LINES FROM MY CLAUDE INSTRUCTIONS AFTER EVERY DRAFT
CAME BACK SOUNDING LIKE A BROCHURE. ALL 5, PLUS WHAT I WROTE
INSTEAD.
5 7,114 HOW I SET UP A CLAUDE PROJECT IN 20 MINUTES.
6 5,687 5 CLAUDE PROMPTS FOR A 60-PAGE DOCUMENT YOU HAVE NO TIME TO READ.
7 5,621 THE 45-MINUTE WEEKLY CLAUDE ROUTINE ANYONE CAN COPY.
8 3,599 10 CLAUDE PROMPTS I RUN ON EVERY DRAFT BEFORE IT GOES OUT.
9 2,829 10 CLAUDE PROMPTS I RUN BEFORE I WRITE A SINGLE WORD OF A DRAFT.
10 2,612 5 THINGS I WISH I HAD BANNED BEFORE I LET CLAUDE WRITE IN MY VOICE.
11 2,445 5 LINES I BANNED FROM EVERY CLAUDE DRAFT
12 2,114 5 QUESTIONS I NEVER ASK CLAUDE ABOUT MY OWN WORK
13 1,999 6 CLAUDE PROMPTS THAT COVER AN ARTICLE FROM HOOK TO CTA.
14 1,922 5 CHATGPT PROMPTS TO REHEARSE A HARD CONVERSATION BEFORE YOU HAVE IT.
15 1,753 6 CHATGPT PROMPTS THAT ARGUE WITH YOUR DRAFT INSTEAD OF PRAISING IT
16 1,703 10 CLAUDE PROMPTS I KEEP IN ONE PROJECT TO PLAN, DRAFT AND TIGHTEN A POST.
17 1,684 5 THINGS I NEVER ASK CHATGPT TO DO
18 1,598 HOW I MAKE CLAUDE DISAGREE WITH ME, IN 5 STEPS.
19 1,438 4 PROMPTS THAT MAKE AI MARK EVERY CLAIM IT MADE UP BEFORE YOU PUBLISH.
20 1,398 THE 4-PART BRIEF I PASTE BEFORE ANY CLAUDE TASK.Read down the list once without stopping. Your eye will find the pattern before I name it.
Almost every line is short. Twenty of the 21 non-swap lines in ranks 4 to 25 run 74 characters or fewer.
Almost every line promises a count. Almost every line names one specific thing and one specific object.
None of them asks the reader a question. None of them starts with a feeling.
Pattern One: The Number Comes First
Sixteen of my top 25 first lines start with a digit. Those lines are 9% of my posts and 16% of my views.
That is about 1.7 times the reach of the average post, from a pattern that costs you three keystrokes.
The number does a quiet job. It tells the reader how long the post is and that it ends.
A reader who sees “5 THINGS” knows the work. They will read five items and stop. A reader who sees an open-ended opener cannot price the effort, so they scroll.
The narrower version is stronger. Lines that open with “5 THINGS” are 4% of my posts and 10% of my views, which is 2.9 times the average.
Use a number you can honestly fill. Five items that are each worth reading beat ten items where the last four are padding.
Pattern Two: A Refusal Beats a Recommendation
Look at the top of the list again. “Never ask”, “remove”, “never paste”, “banned”.
Lines built on a refusal are 7% of my posts and 12% of my views. That is 1.7 times the average, and nine of the top 25.
A recommendation says “do this.” A refusal says “I stopped doing this, and here is why.” The second one carries a small story.
It also answers a worry the reader already has. People are more afraid of a wasted hour than they are hungry for a new trick.
The strongest refusal lines also promise the replacement. Rank 3 does it in the hook itself: what I never paste, and what I paste instead.
A refusal with no replacement is a complaint. A refusal with a replacement is a lesson.
Best Social Media Hooks Name the Object and the Cost
Here is what the ranked lines have in common, once you strip the numbers off.
Each names a concrete object. A 60-page document. A project. A draft. An instruction.
Each names a cost or a moment. Twenty minutes. Forty-five minutes. Before it goes out. Before I write a single word.
Compare that with the weak lines from the same two months. “Which Claude word do you still nod along to without really knowing it?” has no object you can hold and no cost you can feel.
Here is a plain test. If your hook could be about anything, it will be seen by no one.
“5 CLAUDE PROMPTS FOR A 60-PAGE DOCUMENT YOU HAVE NO TIME TO READ” can only be about one situation. A reader in that situation stops. A reader outside it scrolls, and that is fine.
Get R Dilip Kumar’s stories in your inbox
Join Medium for free to get updates from this writer.
Specific hooks lose the wrong readers on purpose. That is how they win the right ones.
What the Weakest Hooks Had in Common
I looked at the opposite end too, because the contrast teaches more than the winners do.
Questions were 4% of my posts and 0% of my views. Not one question made the top 25.
First lines that start with “I” or “My” were 10% of my posts and 2% of my views. One reached the top 25, and it was a “HOW I” line with a number and a time in it.
The pattern is not that “I” is banned. The pattern is that “I” with nothing attached to it is a diary entry, and strangers do not read diaries.
Single posts were 68% of my posts and 6% of my views. None reached the top 25.
I will be careful here. That last number is tangled up with the hook. My threads are also where I put my best material, so I cannot say the hook alone did the work.
What I can say is that the hook has to promise something a single post cannot hold. A count of items is exactly that.
How to Write a Hook: The Six-Slot Template Bank
How to write a hook gets about 5,400 searches a month in the United States. The advice people find is usually a list of formulas with no data behind them.
These six templates come from the lines that earned the most on my account. Fill the brackets with something true and specific to you.
HOOK BANK: six fill-in templates
1. REFUSAL + COUNT
[N] THINGS I NEVER [verb] [tool or place], AND WHAT I [do] INSTEAD.
2. PROMPT LIST + SITUATION
[N] [TOOL] PROMPTS FOR [a specific painful situation].
3. HOW I + TIME
HOW I [achieve a specific outcome] IN [N minutes or N steps].
4. ROUTINE + PERMISSION
THE [N]-[MINUTE or PART] [ROUTINE or BRIEF] ANYONE CAN COPY.
5. BANNED LIST + FAILURE
I BANNED [N] [LINES or WORDS] FROM [place] AFTER [specific failure].
ALL [N], PLUS WHAT I WROTE INSTEAD.
6. QUESTION YOU NEVER ASK
[N] QUESTIONS I NEVER ASK [tool] ABOUT [my own thing].
RULES FOR EVERY ONE
- Number first, or within the first three words.
- One object a reader can picture.
- One cost: a time, a size, or a moment in the process.
- Under 75 characters if you can.
- If the hook promises a replacement, the post must deliver it.Run each template on one real thing you did this week. Not a thing you might do.
A hook like “HOW I [do X] IN [N] MINUTES” only works if you timed it. Time it once, and you will write the line with confidence.
The Ten-Minute Audit Prompt for Your Own Account
My ranking is a starting point. Your audience will rank differently, and the way to know is to run the same check on your posts.
You need an export of your last 200 to 600 posts with views. Any tool that lists your posts with their numbers will do.
Paste the export into Claude or ChatGPT with the prompt below.
Here is a list of my social media posts with views, one per line.
Step 1. Take the first line of each post as its hook.
Step 2. Group the hooks by these features and report each group:
- starts with a digit
- is a question
- starts with "I" or "My"
- contains a refusal word (never, stop, banned, remove, avoid)
- is 75 characters or fewer
- belongs to a multi-part thread versus a single post
Step 3. For each group, give me: the share of my posts, the share of
my total views, and the group's average views divided by the average
views of all posts.
Step 4. List the top 25 hooks by views, then say which groups they
belong to.
Do not give advice yet. Only report the numbers. If a group has fewer
than 8 posts, mark it "too small to trust".The last line matters. A group of three posts can look like a pattern and be a coincidence.
I run the “too small to trust” rule on my own data. My top three posts share one opener, and that group is only six posts. I read its numbers as a hint, not as a rule, and I built the template bank from the larger groups instead.
Once the numbers are back, ask for advice in a second message. Keeping the two steps apart stops the tool from inventing a story before it has finished counting.
Social Media Hooks Examples: One Real Before and After
Here is a real weak hook from the same window, and a rewrite using the bank above.
The weak one is a question about a word. “Which Claude word do you still nod along to without really knowing it?” It is a single post, and it has the two features that did worst for me: a question, and no number.
The rewrite uses template 6 and template 1 together: “5 CLAUDE WORDS I NODDED ALONG TO FOR A YEAR, AND WHAT EACH ONE MEANS.”
Same topic. Same reader. The change is a number up front, a cost (a year of nodding), and a promise of a replacement (what each one means).
I have not posted the rewrite, so I cannot show you a result. I am telling you that on purpose. The point of this example is the edit, and the edit is something you can do to your own weakest hook in two minutes.
Do the same: find your lowest-reach post that deserved better, and write the rewrite before you read another word of advice.
What This Data Cannot Tell You
I want to be plain about the limits, because a hook article that hides its limits is just another formula.
Topic is mixed in. Most of my top posts are about Claude, and 59% of all my first lines mention it. A hook about a topic my audience already follows will beat a clever hook about one they do not.
Time is mixed in. Posts went out on different days at different hours, and my account grew over the two months.
The three largest posts skew every average. That is why I split that group out and why I read the rest of the data without it.
None of this was a controlled test. It is a ranking of what happened, and a ranking is good for choosing what to test next.
The cleanest way to use it is to test one template per week. Write the same post with two different first lines on two different days, and compare the views against your own account’s average, not mine.
The Five Rules I Take From the Ranking
If you only keep one section, keep this one.
Start with a number you can fill honestly. Name one object a reader can picture. Name one cost, in minutes or pages or a moment. Prefer a refusal that comes with a replacement. Keep it short.
Then give the post a body worth the promise. A strong hook on a weak thread only teaches your audience to scroll past you faster.
I think about the best social media hooks in one way now. They are a contract. The first line states what the reader will get, and the rest of the post pays it.
Look at the 20 lines in the ranking again. Each one is a promise small enough to keep in a single thread.
Write your next promise before you write your next post.
Threadeazy is a Threads post creation and scheduling tool. Try it here
