You paid for 3 seconds. Most people waste them.

01What a hook is actually for

A hook gets the right person to stop. And it lets the wrong person keep going. That second half is the part people miss.

A hook that stops everybody is a failure with pretty numbers. You end up paying for the attention of people who were never going to buy. Nothing looks wrong on day 1. Then 3 weeks later your cost per result quietly tells you the truth.

A hook that stops everybody is a failure with good numbers.

So before you judge a hook by how smart it sounds, ask 1 thing. Does the person I want recognise themselves in the first few words?

02The 3 parts every working hook has

Every hook that has worked well for me has all 3 of these. Not 2. All 3.

Here is a hook that has spent a huge amount of money. "Little-known secret about Medicare that could cost you thousands." Look at how it does all 3. It calls out people on Medicare. It hints at a benefit, in this case avoiding a loss, which people feel more sharply than a gain. And it leaves a question hanging. What secret?

A completely different one. "This is for my girlies who need a girl's night soon." Calls out an audience. Hints at a benefit you feel rather than argue with. And it opens a question. What is this?

03Words first, picture second

The hook is the first thing people see, and that means both the picture and the words on it. But the words create the picture. So you always start with the words.

People do this backwards all day long. They make a beautiful image, then write a line to sit on top of it. Now the line has to fit a box. You wanted the box built to carry the line. Write the sentence that stops the person, then build the visual around it.

This is also where the company brain inside an AiOS command center earns its keep. If every call note, every ad that worked and every line a buyer actually said is in 1 place, writing the hook starts from real words instead of a blank page.

04The 3 questions before you write anything

Answer these 3 first. They turn hook writing from a lucky guess into a process you can repeat.

The middle question is the one everyone skips, and it changes everything. Someone who has tried nothing needs a simple promise. Someone who has already tried 4 things needs a reason this one is different, or they will not even register your promise as new information.

053 ways to stand once you know that

Once you know how much your market has already tried, you have 3 positions you can take.

A market that has tried nothing responds to A. A tired market that doubts everybody needs B, because A sounds exactly like the thing that already failed them. C works on both, and it is the easiest to do, which is why ads made for 1 specific group of people work so well now.

06You are joining a conversation, not starting one

A hook grabs attention because it matches the conversation already running in someone's head. You are not starting that conversation. You are walking into the middle of it.

This is why hook writing is mostly a listening job and not a writing job. If you know your buyer thinks "I am 42 and I do not recover from anything the way I used to", you can write a line that sounds like their own thought. If you do not know that, you will write something about wellness solutions and they will scroll.

A real line from an account we run. "The 40s reality check: at 42, recovery stops being about rest. It becomes about knowing exactly which muscle stopped doing its job, and most people never find out." That is somebody's own thought, said out loud. The same listening habit is why more people now ask an AI chatbot who to hire instead of searching, which is its own problem and I wrote about whether a chatbot says your name separately.

Most of my best hooks are lightly edited quotes from people who did not know they were writing an ad.

07How you know if the hook worked

There is 1 number for this. Hook rate. It tells you how many people stopped. It is a different number from hold rate, which tells you whether the rest of the ad kept them.

Those are our numbers on our account. They are not a target for you. The value is in comparing your own ads to each other. A hook that stops half as many people as your best one is a hook problem, no matter what number is on somebody else's screenshot.

08The habits that make it repeatable

Write 10 hooks before you like any of them. The first 3 are always the obvious ones, and the obvious ones are what your competitors wrote too.

Keep a running file of the exact phrases buyers use on calls. Not your summary of what they said. Their words. In 2026 I do this with a Telegram assistant that drops the line straight into the brain, so it is there when I sit down to write. Before that I kept it in a note. Either way, keep it.

And test hooks against each other with everything else held still. If you change the hook, the picture and the offer at once, you have learned nothing, and that is the most common way a testing budget disappears. The cheapest place to test a hook is a normal post, before you pay for it.

09Questions, answered