# B2B ad hooks for LinkedIn and Meta: nine types and the buyer fear each one uses > A hook is the first line or first picture a buyer meets in an ad. Business-to-business (B2B) hooks fall into nine types, such as a scene from the reader's week, a question that exposes a gap, a number about peers or borrowed trust, and each type pulls on one fear or wish. This page shows each type in real LinkedIn and Meta ads, says where the first line gets cut, and gives a three-question test and two prompts. Source: https://nimblet.io/guides/b2b-ad-hooks. Updated 2026-10-23. ## What is a hook on LinkedIn and on Meta? A hook is the first line or first picture a business-to-business (B2B) buyer meets in an ad, before the buyer decides whether to look further. It is one line or one picture. Take one hook per ad, from the creative itself, and use the same rule on LinkedIn and on Meta. Nimblet's own ad reading uses the same rule for images and videos. On both platforms the intro text sits above the image or video, so the reader meets the text and the media almost together. The table shows which line we take as the hook. When the image and the intro text carry different lines, we tag the line from the table and name the other one. | Format | The hook is | |---|---| | Single image ad | The largest text in the image (in an animated image, the first line it builds). If the image holds only a title, a logo or a photo, the first sentence of the intro text or primary text. | | Video ad | The first seconds: on-screen text first, then speech, then the first visual move. | | Document ad | The cover title if it carries the idea. If it only names the file, the first sentence of the intro text. | | Carousel ad | The first card. If the cards only continue the intro text, the first sentence of the intro text. | | Thought leader ad (TLA) or text post | The lines before the "see more" link. | A hook and an angle are two things. The angle is the reason the buyer would act ([B2B ad angles](/guides/b2b-ad-angles)). The hook is the first line that opens that reason, so one angle can open with ten hooks. Proof ([proof points in B2B ads](/guides/proof-points-in-b2b-ads)) and the offer come after the hook, and the lever behind it is in [the psychology of B2B ads](/guides/psychology-of-b2b-ads). This page covers paid ads, not organic posts or profile headlines. For the post behind a thought leader ad, see [LinkedIn thought leader ads](/guides/linkedin-thought-leader-ads). Our reading is that a LinkedIn reader meets an ad in a work frame and a Meta reader in a personal one. [LinkedIn ads vs Facebook ads for B2B](/guides/linkedin-ads-vs-facebook-ads-b2b) follows that difference through the whole ad. ## Which hook types do B2B ads use? B2B ads open with nine types of hook: Scene, Reframe, Question, Number, Borrowed trust, Stakes, Switch, Product move and Offer. We built the list from 61 public ads, 41 from the LinkedIn Ad Library and 20 from the Meta Ad Library. We read them on 22 October 2026 with the country filter set to Germany, which keeps ads that target Germany. They come from software and services brands. We coded each hook ourselves, so the list is our reading and not a standard. Many ads are in German. We quote the original and give our translation. Many ads sell artificial intelligence (AI) features, so the word shows up in many quotes. The libraries show no clicks, spend or conversions for these ads, so nothing below says that an ad worked. We found no public study that ranks hook types in paid B2B ads, so this page ranks nothing. The LinkedIn Ad Library has no thought leader label. We call an ad a thought leader ad when it shows a person as the author and the company as the promoter. Each ad name opens a screenshot of the ad as the library showed it in October 2026. We quote the text the libraries give us. For video ads that is the intro text, because we did not watch the first seconds. Where an image carries a different line from the intro text, we say so. ### Scene The first line puts the reader in a moment from their own week: a chore, a mess or an awkward scene. Nothing is for sale yet. The reader only has to think "that is me". - [Workday, LinkedIn single image ad](/guides/ads/1640400306.jpg): the image is animated and first builds the line "Die ganze Woche nur Kalendern und Postfächern hinterherjagen?" Our translation: the whole week just chasing calendars and inboxes? The intro text opens "Niemand wird Recruiter, um ständig Kalendern und Postfächern hinterherzujagen." (nobody becomes a recruiter to keep chasing calendars and inboxes). A later frame says "Die Terminplanung läuft ab jetzt von selbst." (scheduling now runs by itself). - [Zendesk, LinkedIn single image ad](/guides/ads/1574291156.jpg): the intro text opens "Etwas lief nicht so, wie es sollte. Vielleicht hatten Sie einfach einen schlechten Tag und die Customer Experience (CX) ließ deshalb zu wünschen übrig." Our translation: something did not go as it should. Maybe you just had a bad day and the customer experience (CX) suffered. The image holds only the title of a pack of apology letter templates. - [Celonis, LinkedIn thought leader ad](/guides/ads/1605611964.jpg): "Picture this: you bring in a temp through a staffing agency, and over two years they learn exactly how your business runs". The scene is an analogy for losing control of your own company's know-how. Exit Five shows how one B2B creative director builds a scene. He used Reddit Answers to find posts where human resources (HR) managers describe their challenges. He saw that they call approval back-and-forth "email ping pong". He took the phrase literally. The newsletter sets "the line every HR software ad runs", "Automate time-off requests and approvals", against his version: "An employee asks for a Friday off. Then form B3.2, then Section 4B, then 'there is no Section 4B on my copy'" ([Exit Five, newsletter 271, 28 July 2026](https://exitfive.com/newsletter/b2b-paid-ads-that-dont-suck-proven-creative-recipes-that-sell-anything-even-boring-products/)). Exit Five calls the recipe a visual analogy. The post reports case results that are self-reported, so we cite none. ### Reframe The first line gives the reader a new view of an old problem. It often takes the blame off the reader. - [monday.com, LinkedIn document ad](/guides/ads/1588880964.jpg): the cover and the intro text both say "Dein Team ist nicht zu langsam. Es ist überplant." Our translation: your team is not too slow, it is overplanned. - [HubSpot, Meta video ad](/guides/ads/1136999918997979.jpg): "KI mit Kontext. Endlich." The description adds "Generische KI erzeugt generische Ergebnisse." Our translation: AI with context, at last. Generic AI produces generic results. - [Lexware, LinkedIn thought leader ad](/guides/ads/1572651173.jpg): "Faul sind die Jungen nicht. Sie fragen nur schneller, warum wir etwas eigentlich so machen." Our translation: young people are not lazy, they just ask faster why we do things this way. ### Question The first line asks one role a question. The strong ones expose a gap the reader cannot close in the moment. - [Adobe, LinkedIn thought leader ad](/guides/ads/1584237893.jpg): "Wer ist in eurem Unternehmen eigentlich für KI verantwortlich?" Our translation: who in your company is actually responsible for AI? The next lines ask for a clear mandate, a budget and the power to decide across teams. - [Okta, LinkedIn single image ad](/guides/ads/1629578306.jpg): "Can you name every AI Agent & what they do in your systems?" It is the large line in the image. - [Pipedrive, LinkedIn video ad](/guides/ads/915417333.jpg): "Which should come first: the discovery call or the product demo?" It is a debate question for people who sell. ### Number The first line opens with a figure about the reader's peers or the reader's problem. A number gives the reader something to check. - [Personio, LinkedIn single image ad](/guides/ads/1584971803.jpg): "Fast neun von zehn Unternehmen sind sich sicher: Unsere Lohnabrechnung stimmt. 1.200 befragte Lohnverantwortliche und CFOs zeichnen ein anderes Bild." Our translation: almost nine in ten companies are sure their payroll is correct, and 1,200 payroll owners and chief financial officers who answered a survey paint a different picture. The survey is the advertiser's own. The largest text in the image is the report title, so we take the intro text. - [Lexware, LinkedIn thought leader ad](/guides/ads/1584431403.jpg): "100 Euro gespendet, rund 65 Euro bezahlt." Our translation: 100 euros donated, about 65 euros paid. The next line says that the tax office carries the rest. - [Freshworks, LinkedIn document ad](/guides/ads/1597821666.jpg): "43.6% now use AI across service workflows, up from 32%." The cover names a benchmark report, so the report is the source. ### Borrowed trust The first line is somebody else's word: a named customer result, a customer quote or a third-party badge. [Proof points in B2B ads](/guides/proof-points-in-b2b-ads) sorts what counts as proof. The hook only decides that proof goes first. - [ServiceNow, LinkedIn single image ad](/guides/ads/1577259373.jpg): the image pairs the Lenovo and ServiceNow logos under the line "Lenovo works with ServiceNow". The intro text opens "Lenovo cut case resolution time by 35% with AI agents handling the grind." The ad gives no source for the 35%. - [monday.com, LinkedIn single image ad](/guides/ads/1578354254.jpg): „Das Enterprise-Paket von monday.com erfüllt alle unsere Datenschutz- und Compliance-Vorschriften.“ Our translation: the monday.com Enterprise plan meets all our data protection and compliance rules. The quote is the largest text in the image. The ad names the customer and the speaker's job title. In our reading, the quote answers a doubt that the legal and information technology (IT) teams would raise. - [Atlassian, LinkedIn single image ad](/guides/ads/1644873956.jpg): the image says "A Leader. Again." and carries a Gartner report label. The intro text opens "Atlassian is named a Leader in Collaborative Work Management for the 2nd year running." The ad credits Gartner. - [Miro, LinkedIn single image ad](/guides/ads/1644816336.jpg): the image says "How Proximie turned weeks-long roadmap debates into work that ships in minutes." The intro text opens with a reframe: "Proximie's hardest problem was never engineering the product. It was deciding what to build next without defaulting to the loudest voice in the room." ### Stakes The first line names a loss, a rule or a deadline the reader must answer for. Our pull holds two clear cases. - [Personio, Meta image ad](/guides/ads/1282855173922184.jpg): the image says "Hey HR-Teams! Ist Ihr Unternehmen auf die Transparenzpflicht vorbereitet?" Our translation: hey HR teams, is your company ready for the transparency duty? The headline reads "Entgelttransparenz 2026: Aus Pflicht wird Wettbewerbsvorteil" (pay transparency 2026, duty becomes competitive advantage). The description names the European Union (EU) pay transparency directive. - [Lexware, Meta video ad](/guides/ads/1078108978065275.jpg): "Wenn die Buchhaltung kippt, kippt alles". The text then lists "verpasste Fristen, Mahnungen, unnötiger Stress". Our translation: when bookkeeping tips over, everything tips over, with missed deadlines, payment reminders and needless stress. ### Switch The first line names the old tool, or the reason to leave it. - [HubSpot, LinkedIn single image ad](/guides/ads/1585176223.jpg): the image says "Zeit für ein neues Marketingsystem?" Our translation: time for a new marketing system? The intro text opens "Deine Systeme kosten Zeit und Nerven? Diese Live-Demo ist für Teams, die über einen Wechsel zu HubSpot nachdenken." (do your systems cost you time and nerves? This live demo is for teams that are thinking about switching to HubSpot). - [Shopify, LinkedIn single image ad](/guides/ads/1648714386.jpg): the image leads with "Get the fastest time to value and better TCO" and a large "31% better TCO than BigCommerce". TCO means total cost of ownership. The small print says the data comes from research commissioned by Shopify. The intro text opens "Your competitive advantage became a maintenance burden." The headline names the rival platform again: "See what's possible beyond BigCommerce". - [sevDesk, Meta image ad](/guides/ads/1359588102914002.jpg): the image shows sticky notes that read "Immer noch bei deinem alten Buchhaltungstool?" Our translation: still on your old bookkeeping tool? The primary text opens "Ein altes Tool macht deine Buchhaltung unnötig kompliziert." (an old tool makes your bookkeeping needlessly complicated). - [monday.com, LinkedIn single image ad](/guides/ads/1578255844.jpg): the image says "Deine Projekte sind zu groß für Tabellen." Our translation: your projects are too big for spreadsheets. The headline reads "Board statt Tabelle" (board instead of spreadsheet). ### Product move The first line says what the product does, or the first frame shows it doing the job. - [Zendesk, LinkedIn video ad](/guides/ads/1546752236.jpg): "Unser intelligenter KI-Assistent kann mehr als nur beraten." Our translation: our intelligent AI assistant can do more than advise. - [HubSpot, Meta image ad](/guides/ads/947865284646859.jpg): the image says "Stellen Sie ein Team aus KI-Agents zusammen, das die Dinge für Sie erledigt." Our translation: build a team of AI agents that gets things done for you. The primary text says "Beschreiben Sie die Aufgabe, und der Agent-Builder erstellt einen KI-Agent, der sie für Sie erledigt, ganz ohne Code." (describe the task, and the agent builder creates an AI agent that does it for you, with no code at all). - [HubSpot, Meta image ad](/guides/ads/2025445024835130.jpg): the image says "Meeting-Notizen erfasst, Deals aktualisiert. Automatisch." Our translation: meeting notes captured, deals updated, automatically. The headline says "Nie wieder Gespräche rekonstruieren" (never reconstruct conversations again). The primary text says that Notetaker records the call, writes the summary and drafts your next step. ### Offer The first line names what the reader gets: a template, a report, a live session or a discount. - [Personio, Meta image ad](/guides/ads/2330341690707775.jpg): the image says "50% weniger Kosten. 100 % HR-Exzellenz." Our translation: 50% lower cost, 100% HR excellence. The small print says that non-profit organisations get Personio at half the price. The headline reads "50% Nachlass & Ihr individuelles Expertengespräch" (50% discount and your personal expert call). - [Snowflake, LinkedIn document ad](/guides/ads/1633878686.jpg): "Snowflake's leadership share the exact framework behind their GTM AI transformation, including what didn't work. Free ebook." GTM means go-to-market. The cover only names the ebook, so we take the intro text. - [Salesforce Developers, LinkedIn event ad](/guides/ads/1648734276.jpg): "AIforce AMA: Build Rich Agentic UI with Headless Experience Layer". AMA means ask me anything. The text promises that experts answer your questions live. ### How does this list compare with others? Most ad tool blogs we read in October 2026 define a hook as the first line or the first 3 seconds. Most of them cover consumer ads on Meta and TikTok and do not discuss B2B, LinkedIn or buying groups. That advice leaves out the reader's buying role and the fear a hook uses, so this page adds both. It also reads real ads, and it does not say that any hook type wins. ## Which buyer fear or wish does each hook type use? Each hook type pulls on one fear or one wish, and the type should follow the fear of the reader you chose. The fear column is the core of this table. It is our reading and we did not test it. The awareness stages are the five from Eugene Schwartz's book Breakthrough Advertising: unaware, problem-aware, solution-aware, product-aware and most-aware. Pairing a type with a stage is also our reading. | Hook type | What the first line does | Fear or wish it uses | Awareness stage it suits | Real example | How it fails | |---|---|---|---|---|---| | Scene | Puts the reader in a moment from their own week | Wish: relief from a chore. Fear: being the one who lives with the mess. | Unaware, problem-aware | [Workday](/guides/ads/1640400306.jpg) | The scene fits every category, so a rival can run it. Use the reader's own word for the pain. | | Reframe | Gives a new view of an old problem | Wish: to be seen as right. Fear: blame for a problem the reader did not cause. | Problem-aware, solution-aware | [monday.com](/guides/ads/1588880964.jpg) | Nothing sits behind the new view, or the line blames the reader. | | Question | Asks one role a question that exposes a gap | Fear: having no answer when a boss, an auditor or a peer asks. | Problem-aware | [Adobe](/guides/ads/1584237893.jpg) | The question has one possible answer, or it names no role. | | Number | Opens with a figure about peers or the problem | Fear: falling behind peers. Wish: to know where I stand. | Problem-aware, solution-aware | [Personio](/guides/ads/1584971803.jpg) | The ad gives no source, or the number measures nothing the reader owns. | | Borrowed trust | Opens with a named customer result, a customer quote or a third-party badge | Fear: a choice nobody can defend later. Wish: the safe pick that people like me made. | Product-aware | [monday.com](/guides/ads/1578354254.jpg) | The customer has no name, the result is a promise, or the badge means nothing to the buying group. | | Stakes | Names a loss, a rule or a deadline | Fear: a penalty, an audit or a missed date. Wish: to be the one who saw it coming. | Unaware, problem-aware | [Personio](/guides/ads/1282855173922184.jpg) | The rule is stated as fact with no date or source, and the ad names no next step. | | Switch | Names the old tool, or the reason to leave it | Fear: staying stuck with the old tool. Second fear: the cost of leaving. | Solution-aware | [HubSpot](/guides/ads/1585176223.jpg) | It names the pain of staying and says nothing about the cost of leaving. | | Product move | Says what the product does, or shows it doing the job | Wish: a result with less work. Fear: buying a tool that adds work. | Solution-aware, product-aware | [HubSpot](/guides/ads/947865284646859.jpg) | It names a capability the reader never asked for, or says "AI" and names no task. | | Offer | Opens with what the reader gets | Wish: a small, safe first win. Fear: wasting time or budget. | Problem-aware for a guide, most-aware for a discount or a demo | [Personio](/guides/ads/2330341690707775.jpg) | It attracts people who want the free thing and not the product. | The outside evidence we found is about how B2B buyers decide. LinkedIn's Buyability study asked 750 senior B2B buyers what drives a decision. [The Drum reports](https://www.thedrum.com/news/2025/06/18/linkedin-s-new-b2b-playbook-shows-ability-defend-bad-decisions-drives-b2b-buying) two findings. The ability to defend the decision if it goes wrong was the most important of five emotional dimensions. Recommendations from "people like me" were the single most influential factor. This is LinkedIn's own research and we read the coverage, not the study. Our reading is that Borrowed trust, Question and Stakes speak to defending a choice. We did not test that. A hook that names one role speaks to one member of a buying group. [Gartner](https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process) reports buying groups of five to 16 people across as many as four functions (survey of 632 buyers, August to September 2024). It found that content tailored to the buying group raised consensus, and content tailored to one person's view lowered it. The survey was about sales content, not ads. Our reading: name one role in the hook, but let a colleague in finance or IT repeat the line without an argument. Pick the type by what you hold: - You hold a sourced customer result or quote: Borrowed trust. - You hold your own survey data: Number. - Your reader uses a spreadsheet or an old tool: Switch. - A rule or a date applies to your reader: Stakes. - Your product removes a chore the reader names: Product move. - You hold a template, a report or a live session: Offer. - You hold none of these: Scene, Reframe or Question. They need only a true insight about the reader's week. ## Where does the fold cut, per format? LinkedIn says to keep intro text to 150 characters to avoid truncation on single image, video and document ads, and gives the same figure for carousel ads. Meta recommends 125 characters of primary text on Instagram Feed and 50 to 150 on Facebook Feed, and does not say where the cut falls. We found no official number for a thought leader post. In the library previews we opened, the cut fell after three lines whatever the character count, so check the preview on a phone. | Format | Where the hook sits | Official figure we read | Source | |---|---|---|---| | LinkedIn single image | Image text, then intro text | Intro text 150 characters and headline 70 characters to avoid truncation | [LinkedIn help](https://www.linkedin.com/help/lms/answer/a426534) | | LinkedIn video | First seconds, then intro text | Intro text 150 and headline 70 characters. LinkedIn says to show what you want the audience to see in the first 10 seconds and that many members watch with the sound off. | [LinkedIn help](https://www.linkedin.com/help/lms/answer/a424737), [video tips](https://business.linkedin.com/marketing-solutions/success/best-practices/video-ad-tips) | | LinkedIn document | Cover title, then intro text | Intro text 150 and headline 70 characters. The headline defaults to the file name. | [LinkedIn help](https://www.linkedin.com/help/lms/answer/a493903) | | LinkedIn carousel | First card, then intro text | Intro text 150 characters (maximum 255). A card headline shows two lines before it is cut. | [LinkedIn help](https://www.linkedin.com/help/lms/answer/a423087) | | LinkedIn thought leader post | The lines before "see more" | No official number found | See below | | Meta Facebook Feed image | Image text, then primary text | Primary text 50 to 150 characters recommended, headline 27 | [Meta ads guide](https://www.facebook.com/business/ads-guide/update/image/facebook-feed/link-clicks) | | Meta Instagram Feed image | Image text, then primary text | Primary text 125 characters, headline 40 | [Meta ads guide](https://www.facebook.com/business/ads-guide/update/image/instagram-feed) | | Meta video | First seconds | Captions and sound are optional but recommended | [Meta ads guide](https://www.facebook.com/business/ads-guide/update/video/facebook-feed) | Four details change where the hook can live. - On Meta, an Awareness objective hides the headline, description and button of a video ad on Facebook Feed on mobile. So do some reach and video-view goals ([ads guide for video in Facebook Feed](https://www.facebook.com/business/ads-guide/update/video/facebook-feed)). A hook that lives in the headline is invisible there. - A LinkedIn document ad takes its headline from the file name unless you change it. Name the file as the hook. - We found no official fold for thought leader posts. [Typefully's previewer](https://typefully.com/tools/linkedin-post-preview) describes the cut in lines, about three to five on desktop and tighter on mobile. A [contentin.io post](https://contentin.io/blog/linkedin-post-character-limit/) (updated 22 October 2026) says LinkedIn does not document the point and that it varies by device and screen width. We print no official number. - Jon Loomer's table lists 125 characters for Facebook Feed primary text and says the text is truncated after that ([Jon Loomer](https://www.jonloomer.com/facebook-ad-dimensions-and-character-limits/), page modified 1 March 2025). Meta's own page calls its figures recommendations. We also looked at how the cut falls in practice. We opened two LinkedIn single image ads in the Ad Library, [Personio](/guides/ads/1584971803.jpg) and [Okta](/guides/ads/1629578306.jpg), in a desktop-width window (1,280 pixels) and in a phone-width window (390 pixels). Both previews stopped after three lines. At desktop width that was about 210 and 220 characters. At phone width it was about 110 and 120. The count moves by a few characters because the "see more" link covers the end of the last line. In the thought leader posts we opened (Adobe, Celonis and Lexware), the preview also stopped after three lines, and a blank line counted as a line. At phone width, Adobe's two-line question was followed by a blank line, so the preview ended right after the question. This is the library's preview, not the live feed, and the library says versions can look different by device. It suggests that not all 150 characters show on a phone. In our reading, put the hook in the first 100 characters. For video, ad tool vendors often say the first 1 to 3 seconds decide whether someone keeps watching. We found no study behind that claim. Hook rate is usually defined as 3-second video plays divided by impressions. Vendor pages that give a range for a solid hook rate cite no source, so we print no benchmark. To check your own ads, open the ad preview in LinkedIn Campaign Manager and in Meta Ads Manager. Switch the device or the placement. Count the characters you see before "see more", and write the number into your brief. ## How do you test that a hook is specific? Run three questions on every hook. Can the reader picture it? Can someone prove it wrong? Can nobody else say it? They come from Harry Dry's copy rules, as [Demand Curve's growth newsletter 202](https://www.demandcurve.com/newsletters/growth-newsletter-202) reports them (8 August 2024). Dry's rule of thumb: three no answers mean you have probably written rubbish, and three yes answers mean you are onto something. His examples are consumer products, so here are B2B ones. We scored four real hooks. This scores the line, not the ad, and it is our reading. | Hook | Picture it? | Prove it wrong? | Only this advertiser? | Reading | |---|---|---|---|---| | [Workday](/guides/ads/1640400306.jpg): the whole week just chasing calendars and inboxes? | Yes | No | No | Calendars and inboxes are concrete. A feeling cannot be proven wrong, and any scheduling vendor could ask the question. | | [Personio](/guides/ads/1584971803.jpg): nine in ten are sure, 1,200 surveyed say otherwise | No | Yes | Yes | The survey can be checked and only Personio ran it. Numbers do not make a picture. | | [Lexware](/guides/ads/1584431403.jpg): "100 Euro gespendet, rund 65 Euro bezahlt." (100 euros donated, about 65 paid) | Yes | Yes | No | Two sums the reader can check with a calculator. The tax rule is public, so any tax tool could tell it. | | [Miro](/guides/ads/1644816336.jpg): how Proximie turned weeks-long roadmap debates into work that ships in minutes | No | Yes | Yes | The named customer can confirm or deny the story, and only Miro and Proximie can tell it. Debates and minutes do not make a picture. | None of these four passes all three tests. The two hooks that only their advertiser can say are hard to picture. The two hooks that are easy to picture are lines a rival could run. In our reading, the picture test and the only-us test pull against each other, so look for one concrete detail that only you own. Here is how a weak line changes under the three tests. Approvely is an invented product that collects invoice approvals in one place. The lines are constructed, not tested ads. | Line (constructed) | Picture | Prove wrong | Only us | Fix | |---|---|---|---|---| | Take control of your invoice approvals. | No | No | No | Start from a scene the reader knows. | | Reminder number four. Same invoice. | Yes | No | No | Add a detail that only your product has. | | Managers approve from the reminder email. No login. | Yes | Yes | Only if no rival does the same | Check the claim on your product page before design. | The test does not check the reader or the angle. A hook can pass all three and still talk to the wrong person, so choose the reader first ([B2B creative brief template](/guides/b2b-creative-brief-template)). A hook that states a fact also needs a source in the ad or on the landing page. Copy the card below into your team doc, one card per hook. ```template HOOK CARD Ad and format: [LinkedIn single image / LinkedIn video / Meta image / ...] Reader: [one role at one kind of company] Angle: [one sentence: the reason this reader would act] Hook: [the exact first line or first picture] Where it sits: [image text / first seconds / intro text / cover page / lines before see more] Characters: [count] of [limit from the ad preview] Type: [Scene / Reframe / Question / Number / Borrowed trust / Stakes / Switch / Product move / Offer] Fear or wish: [one, in the reader's words] Test 1, can the reader picture it? [yes / no, and what they picture] Test 2, can someone prove it wrong? [yes / no, and how] Test 3, can nobody else say it? [yes / no, and who else could] Source needed: [the claims in the hook that need a named source, or "none"] What the next ad changes: [the hook only, or the angle] ``` ## How do you write ten hooks for one angle? Give the model one angle, one reader and the nine types, and make it write ten hooks across at least six types. Then cut the ten to three by the three tests, by fear and by claim risk. Writing ten forces a spread. Writing three gives you three versions of the first idea. Scenes and questions rarely pass the prove-it-wrong test, so the prompt drops only a hook that fails all three tests. The prompt below works when pasted alone. Fill every bracket, and write "none" where you hold nothing. ```prompt You write the first line of B2B ads for LinkedIn and Meta. Write ten different hooks for ONE angle. A hook is the first line or first picture the reader meets, before they decide to look further. MY INPUT Product: [what it is and what it does, in one or two sentences] Reader: [one role at one kind of company] What this reader fears or wants, in their words: [the fear or wish; paste phrases from sales calls, reviews or support messages if you have them] Angle: [the one reason this reader would act, in one sentence] Proof I can show: [a named customer result, a quote, or a number with its source; or "none"] Offer and next step: [what the reader gets, and what they do next] Format: [LinkedIn single image / LinkedIn video / LinkedIn document / LinkedIn thought leader post / Meta image / Meta video] Where the hook sits: [text in the image / first seconds of the video / intro text / cover page / lines before see more] Character limit for the hook: [a number from the ad preview, or "none"] Hooks already run or rejected: [list, or "none"] Competitor hooks to avoid echoing: [list, or "none"] THE NINE HOOK TYPES Scene: puts the reader in a moment from their own week, such as a chore or a mess. Nothing is sold yet. Reframe: gives a new view of an old problem. It often takes blame off the reader. Question: asks one role a question that exposes a gap the reader cannot close in the moment. Number: opens with a figure about the reader's peers or the reader's problem. Borrowed trust: opens with a named customer result, a customer quote or a third-party badge. Stakes: names a loss, a rule or a deadline the reader must answer for. Switch: names the old tool, or the reason to leave it. Product move: says what the product does, or shows it doing the job. Offer: opens with what the reader gets, such as a template, a report, a live session or a discount. RULES 1. Write ten hooks. Use at least six different types. Use no type more than twice. 2. Every hook uses one fear or one wish of this reader. Name it. 3. Test each hook before you show it with three questions: can the reader picture it, can someone prove it wrong, can nobody else say it. If a hook fails all three, rewrite it first. Use the answers when you pick the strongest and the weakest hook. 4. Never invent a number, customer, quote, study, award or result. If a hook needs one and I gave none, write a placeholder in square brackets, such as [number from your source], and list it under NEEDS A SOURCE. 5. Use the reader's own words when I gave them. 6. Name an enemy that is a thing, such as an inbox, a spreadsheet or a process. Never blame the reader or a colleague. 7. Make no legal or compliance promise, such as "compliant" or "audit-proof", unless my proof field contains it. 8. Use plain words and active verbs. Write short sentences. Use no dashes, no hype words and no "not X but Y" lines. 9. Each hook fits the character limit. Count spaces. If you are not sure of a count, write "about". 10. For video, write the first on-screen line and the first spoken line, and say what the first frame shows. OUTPUT First, a table with these columns: number, type, hook, fear or wish it uses, characters, needs a source (yes or no). Then three lines: the strongest hook and why, the weakest hook and what would fix it, and the one hook you would test first. Then the list NEEDS A SOURCE, with the claim each placeholder stands for. Write nothing else. ``` Models miscount characters. Paste the hooks you keep into the ad preview and count there. Models also invent proof, which is why rule 4 forces a placeholder. For the full AI workflow, from context to review, see [writing B2B ads with Claude or ChatGPT](/guides/write-b2b-ads-with-claude-or-chatgpt). Here is an illustrative output. Approvely is an invented product. The angle is "the audit question" from the [brief template](/guides/b2b-creative-brief-template). The format is a LinkedIn single image with the hook in the image text, and the limit is 100 characters. These are constructed lines, not tested ads. | # | Type | Hook | Fear or wish | Characters | Needs a source | |---|---|---|---|---|---| | 1 | Scene | The auditor asks who approved invoice 4471. You open your inbox. | Fear: no answer when the auditor asks | 64 | No | | 2 | Question | Can you show who approved your last invoice without opening email? | Fear: no answer when the auditor asks | 66 | No | | 3 | Number | [Number from your own survey] of accounts payable teams answer auditors from their inbox. | Fear: falling behind peers | about 65 once filled | Yes | | 4 | Borrowed trust | [Customer quote about answering the auditor, with name and role] | Wish: the safe pick others made | depends on the quote | Yes | | 5 | Reframe | Your managers do approve invoices. Their answers sit in email. | Fear: no answer when the auditor asks | 62 | No | | 6 | Stakes | Audit season starts in [month]. Your approvals are still in email. | Fear: a missed date | about 65 once filled | Yes, the month | | 7 | Switch | Keep your finance system. Drop the email chase. | Fear: a rollout managers ignore | 47 | Yes, confirm it runs next to the system | | 8 | Product move | Every invoice, its approver and the date, on one screen. | Wish: a result with less work | 56 | Yes, confirm the screen exists | | 9 | Offer | Free checklist: the questions an auditor asks about invoice approvals. | Wish: a safe first win | 70 | No | | 10 | Scene | Four inboxes and a spreadsheet. The auditor wants one name. | Fear: no answer when the auditor asks | 59 | No | Our reading of this output: hooks 1, 2, 5, 9 and 10 need no source, so they can go to design. Hooks 3, 4, 6, 7 and 8 stay as drafts until a person fills the bracket or confirms the claim. Hook 9 is the safest ask and the weakest picture. A good first test is hooks 1, 2 and 5, because they differ in type and share one fear. That isolates the type from the fear. How to cut and judge the result is in [share creative learnings](/guides/share-creative-learnings). Put the hook type into the ad name, for example `2026-10_LI_audit-question_scene_v1`, so you can group results by type later. The naming rule is in the same guide. ## How do you read a competitor's hook? Take one hook per ad, tag its type and fear, count unique hooks and not library entries, and never call a hook a winner. The LinkedIn and Meta ad libraries show the ad, the format and the advertiser. They show no clicks, spend or results. For ads delivered in the EU they add run dates, targeting and an impressions or reach figure. Run length is a weak hint at best. [LinkedIn's help page](https://www.linkedin.com/help/linkedin/answer/a1517918) lists what the Ad Library holds: the ad preview, format, advertiser and payer, and for ads targeted to the EU the impressions, targeting and dates. [Meta's page](https://transparency.meta.com/researchtools/ad-library-tools/) says it offers more information for ads that deliver an impression in the EU, and shows spend and reach for ads about social issues, elections or politics. [LinkedIn ad library](/guides/linkedin-ad-library) covers searching both libraries and the full competitor sheet. Here is the hook part. 1. Pull the ads and open each one. 2. Pick one hook per ad with the rule from the first section, and quote it exactly, in its own language. 3. Count unique hooks, not entries. In our pull, one payroll software advertiser's first ten LinkedIn results were ten separate entries built from three body texts and four headlines. Meta cards say how many ads share a creative and text. 4. Give each hook a type and name the fear or wish. Name whose fear it is. 5. Count hooks per type across the set. A type with no ads raises a question. It may be open space, or a type the market tried and dropped. The library cannot say which. 6. Do not call a hook a winner or a loser from run dates. Here is the method on four Meta ads from one advertiser, read on 22 October 2026. The library shows no results for them, so this reads hooks, not performance. The two videos show their primary text here, because we did not watch the first seconds. | Ad | Hook | Type | Fear or wish | |---|---|---|---| | [Personio, payday (video)](/guides/ads/4533679516958156.jpg) | "Zahltag. Für viele der Höhepunkt im Monat" (payday, for many the high point of the month). The line then turns to payroll owners and "ein echter Stresstest" (a real stress test). | Scene | Fear: payroll goes wrong on the day everyone watches | | [Personio, podcast episode (video)](/guides/ads/1640890307816366.jpg) | "Payroll-Fehler passieren. Aber ein fremder Bonus auf dem Konto der Kollegin?" (payroll errors happen, but someone else's bonus on a colleague's account?) | Scene | Fear: a public mistake in front of colleagues | | [Personio, pay transparency webinar (image)](/guides/ads/1282855173922184.jpg) | "Hey HR-Teams! Ist Ihr Unternehmen auf die Transparenzpflicht vorbereitet?" (hey HR teams, is your company ready for the transparency duty?) | Stakes | Fear: a missed legal duty. Wish: to turn it into an edge. | | [Personio, half price (image)](/guides/ads/2330341690707775.jpg) | "50% weniger Kosten. 100 % HR-Exzellenz." (50% lower cost, 100% HR excellence). The small print limits the price to non-profit organisations. | Offer | Wish: a lower price, for a non-profit | Our reading: four ads use three types. The two payroll ads open with a scene, one about payday stress and one about a wrong bonus. The legal ad opens with stakes and the discount ad opens with the offer. The legal line is also a question to one role. When a hook fits two types, tag the type that carries the fear, which here is the duty. The set holds no Number or Borrowed trust hook, and we cannot say whether Personio tried them elsewhere. A second set shows the same method. On 22 October 2026 HubSpot's German Meta page showed eight active ads that started the same day, each with its own headline. We coded three of them: [calls](/guides/ads/2025445024835130.jpg), [records](/guides/ads/2351700862305854.jpg) and [agents](/guides/ads/947865284646859.jpg). The calls and agents ads lead with a Product move, each for a different job. The records ad leads with a five-star review quote in the image, which is a Borrowed trust hook, and its primary text then states a Product move. Eight headlines on one day is visible copy testing, not proof of a test design. To code a batch, paste the ads into the prompt below. It works alone. Write the image text yourself if you cannot copy it. ```prompt I paste ads from the LinkedIn Ad Library and the Meta Ad Library. Read only the hook of each ad and code it. Do not judge whether an ad works. The libraries show no clicks, spend or results. For each ad I give: a number, the platform, the format, the library link, the text in the image or the first seconds of the video (if I could read it), the intro text or primary text, and the headline. STEP 1. Pick one hook per ad. Take it from the creative first: the text the eye reads first in an image (in an animated image, the first line it builds), the first seconds of a video, the cover title of a document. If the image holds only a title, a logo or a photo, take the first sentence of the intro text or primary text. Quote the hook exactly, in its original language. Add an English translation if it is not English. STEP 2. Give the hook one type from these nine. Scene: puts the reader in a moment from their own week. Reframe: gives a new view of an old problem. Question: asks one role a question that exposes a gap. Number: opens with a figure about peers or the problem. Borrowed trust: opens with a named customer result, a customer quote or a third-party badge. Stakes: names a loss, a rule or a deadline. Switch: names the old tool, or the reason to leave it. Product move: says what the product does, or shows it doing the job. Offer: opens with what the reader gets. STEP 3. Name the fear or the wish the hook uses, in a few words, and the role whose fear it is. STEP 4. Note where the hook sits: image text, first seconds, intro text, cover title, lines before see more, or headline. STEP 5. Count unique hooks, not library entries. Entries with the same words count once. OUTPUT A table with one row per unique hook and these columns: number, advertiser, platform and format, hook in its own words, translation, type, fear or wish, role, where it sits, entries that share it. Then a count of hooks per type, and the types that appear zero times. Then three sentences on what the set shows. Say "the ads I pasted", never "the market". Never call a hook a winner or a loser. Use no dashes. MY ADS [paste the ads here, one block per ad, in this form: 1 | LinkedIn | single image | link | text in the image | intro text | headline] ``` If you keep the result in a sheet, add a column for the date you read each ad. Libraries change, and a hook you read today may be gone next month. ## Nimblet runs this check Nimblet reads each ad with recent spend in your LinkedIn and Meta accounts and records its hook: the exact words or visual move, and where it sits (on screen, spoken, in the copy or visual). It reads the ads of the competitors you track the same way, so you can set hooks side by side and tag the hook type in the ad name or with a tag rule. Nimblet only reads your ad accounts and never changes a campaign. ## Frequently asked questions ### How long should a B2B ad hook be? Short enough to sit above the fold on a phone. LinkedIn says 150 characters of intro text avoid truncation. In two library previews at phone width we saw about 110 and 120 characters, so in our reading the hook should sit in the first 100. For an image hook, the limit is the space in the picture. Check the preview in Campaign Manager or Ads Manager. ### Is the headline the hook? Not by default. The hook is the first thing the reader meets. On a LinkedIn single image ad that is the largest text in the image. When the image holds only a title, a logo or a photo, it is the first sentence of the intro text. On a Meta video ad with an Awareness objective, the headline is hidden on Facebook Feed on mobile, according to Meta's ads guide, so the hook cannot live there. ### Should the hook name the product? It depends on how much the reader knows. A reader in the unaware or problem-aware stage meets a Scene, a Reframe, a Question or a Number best. A reader who already knows the product can take a Borrowed trust, Product move or Offer hook. The pairing of type and stage is our reading, built on the five awareness stages from Eugene Schwartz. ### Do numbers make a better hook? We found no public study that ranks Number hooks against other types in paid B2B ads. A number gives the reader something to check, so a number with no source in the ad works against the hook, in our reading. In our pull, the numbers in the ServiceNow and Okta ads carry no source in the ad. ### Is a question a good hook? It depends on the question. A question that names a role and exposes a gap, such as Adobe's "who in your company is actually responsible for AI?", gives the reader something to answer. A question with one possible answer says nothing. We found no study of question hooks in paid B2B ads. ### What is a good hook rate? We print none. Hook rate is a video measure: 3-second plays divided by impressions. The vendor pages that give a range for a solid rate cite no source. In our reading, compare each video hook with your own recent videos on the same placement, and judge on the goal the campaign buys. ### Can one hook run on LinkedIn and on Meta? Keep the angle and rewrite the hook for the placement. The fold differs and so does the reader's moment. Check the character limit of each placement, and see [LinkedIn ads vs Facebook ads for B2B](/guides/linkedin-ads-vs-facebook-ads-b2b) for what else changes. ### How many hooks should one angle get? Write ten and keep three. Our [brief template](/guides/b2b-creative-brief-template) asks for three hooks per angle. That number is a choice we made, not a researched figure. Three hooks give a designer a choice without a flood, and ten make you spread across types before you cut. ### Can AI write the hooks? Yes, with two limits. Models invent numbers and customers, so the prompt forces a placeholder. Models also miscount characters, so count in the ad preview. Read the result as a stranger would before design starts ([the full workflow](/guides/write-b2b-ads-with-claude-or-chatgpt)). ### Can Nimblet find the hooks in my account and in my competitors' ads? Yes, on the paid plans, for ads in your connected LinkedIn and Meta accounts and for the competitors you track from the ad libraries. Nimblet reads each ad with recent spend and records its hook, the exact words or the visual move, and where it sits. It ranks your ads against ads bought for the same goal, so you can set the hooks of your best ads side by side. Nimblet records the hook, not its type, so you tag the type yourself, for example in the ad name. Nimblet can then turn what it reads into briefs and draft ads in your brand. It only reads your ad accounts and never changes a campaign. See [creative ideation](/features/ai-ad-generator) and [pricing](/pricing).