Help a B2B marketer plan a paid LinkedIn image-carousel ad. Return a storyboard for review. Do not present it as finished artwork or a launched ad. Everything needed to draft the storyboard is below; the marketer does not need to give you an article, a browsing session or access to reference ads.

BUSINESS INPUTS (replace each bracket with a supplied answer or the literal word Unknown)
- Company, product and offer: [ ]
- Buyer and current situation: [ ]
- One action wanted after seeing the ad: [ ]
- What the destination actually delivers: [ ]
- Exact approved landing URL(s) or Lead Gen Form ID, and destination mode (one landing page, different landing pages, or one form): [ ]
- Approved product facts and their source/owner: [ ]
- Approved customer examples, names, numbers and publication permission, if any: [ ]
- Available image assets/screenshots and rights/owner: [ ]
- Product/claim reviewer and campaign/link/tracking reviewer: [ ]
- Tracking plan: card clicks, landing/form event, qualified CRM outcome, reporting window and attribution limits: [ ]
- Brand constraints and any desired card count: [ ]

METHOD AND FORMAT CHOICE
1. State what the buyer needs to do and the one next action the ad asks them to take. Choose the format before writing cards. A paid image carousel fits separate questions, stages, examples or findings. Choose a Document Ad when the buyer needs to read a continuous guide inside LinkedIn, a video when motion or spoken explanation is part of the evidence, or a single image when there is only one point. If another format fits better, explain the choice and stop before the carousel storyboard.
2. If an image carousel fits, use 2–10 still-image cards. Give each card a distinct job; treat any requested count as a ceiling. Card 1 makes a complete promise to someone who never swipes. Each middle card adds one different question, step, example or finding. The final CTA continues the task begun on card 1. Check that any number promised on the cover matches the content cards.
3. Plan square 1080 × 1080 still images; inspect the actual crop and text at feed size. LinkedIn's specifications checked September 26, 2026 say 2–10 still-image cards, JPG/PNG/nonanimated GIF, up to 10 MB per card, 45-character card headlines, 255-character introductory text (LinkedIn recommends 150 or fewer), and destination URLs beginning http:// or https://, up to 2,000 characters. Recheck current Campaign Manager specs before saving; LinkedIn says saved carousel cards cannot be edited. Sources: https://www.linkedin.com/help/lms/answer/a427022 and https://www.linkedin.com/help/lms/answer/a421188.
4. Choose one destination mode. With landing pages, cards may share one URL or have distinct URLs only when each page fulfills that card's promise. With a Lead Gen Form, every card uses the same form CTA. Never fabricate a URL or form ID. Show the exact destination on every card, or Unknown with its owner.
5. Write ordinary, specific words. For each card, give its job, exact on-card copy, image to create or show, headline of 45 characters or fewer, exact destination/form, claim and source/permission, and owner/review status/open question. Label every fictional layout and mockup. Never invent proof, product capabilities, screenshots, customer names, quotations, results, permissions or accessible links. A public ad shows a structure the advertiser chose; its creative does not tell you whether the ad performed well.
6. Keep measurement layers separate: per-card clicks are interactions, landing-page visits or form events are next actions, and qualified CRM outcomes require later qualification and attribution. If attribution or event tracking is Unknown, say so. Do not turn clicks or swipes into pipeline.

PUBLIC REFERENCE SUMMARY (structure only; do not copy words, visuals, brands, statistics or claims)
- Fleetio, https://www.linkedin.com/ad-library/detail/1428305433: nine captured creative cards use a numbered checklist for government fleet software buyers. The structure is cover, one buying question per card, then a deeper checklist invitation.
- PostHog, https://www.linkedin.com/ad-library/detail/1515629073: seven captured creative cards explain a marketing attribution workflow, moving from parts to steps to a product close. Borrow the one-step-per-card logic only.
- Microsoft Advertising, https://www.linkedin.com/ad-library/detail/1598598776: six captured creative cards move from an AI-shopping claim through figures with source labels to a playbook offer. The figures are advertiser claims, not evidence for your business.
- Capgemini, https://www.linkedin.com/ad-library/detail/1531363983: eight captured creative cards build from curiosity to a thesis, customer-story cards and an event close. Customer names require your own permission if adapted.
- Attentive, https://www.linkedin.com/ad-library/detail/1547300933: eight captured creative cards show message examples and a longer resource, but the advertised count and visible sequence do not fully agree. Match your own cover promise to your actual cards.
- Medius, https://www.linkedin.com/ad-library/detail/1543003393: seven captured creative cards turn a finance report into method, findings, synthesis and a report invitation. Any new finding needs its own real report source.
These are public LinkedIn Ad Library records whose full archived creative sequences were inspected September 2026. Their card destination URLs and performance were not available in the archived records. Do not infer landing-page setup, cost, click rate, leads or revenue from them.

WORKED ADAPTATION, CLEARLY FICTIONAL
Imagine a security software company offering a Vendor Access Review Worksheet to an IT lead at a midsize manufacturer. The buyer would use it to review one current vendor account. A five-card draft opens with “Three questions before you renew vendor access.” The next three cards ask what systems the vendor can reach, who approved access, and when it expires. Card 5 invites the buyer to use the worksheet. These are editorial questions, not product or risk-reduction claims. The worksheet URL, real cover image, product facts, tracking and named approvers are Unknown, so launch status is Hold. Keep this fictional company out of the user's business facts.

MISSING-INPUT RULE (OVERRIDES THE NORMAL OUTPUT)
- If company/offer, buyer/situation or one desired action is absent or Unknown, stop. Output only a short “Missing core inputs” section with up to three focused questions and the fields that cannot yet be drafted. Do not output cards, placeholder copy or Ready status.
- If destination behavior/delivery is absent, mark destination mode and every card destination Unknown; name the campaign owner and keep launch on Hold. Do not imply that a page or form exists. If the offer itself cannot be explained without an absent source, stop and ask for that source as a core input.
- If proof, asset, URL, permission, tracking definition or reviewer is absent, write Unknown in the relevant campaign and card fields, name the owner who must supply it, and draft only source-independent editorial questions or approved facts. Remove any evidence-dependent claim. Never silently substitute a fictional example or reference-ad claim.
- Unknown in a launch-critical field always means Hold. A complete draft is not a launch approval. Do not call any artifact, URL, event, person or source verified unless the user supplied proof of verification.

OUTPUT CONTRACT (ONLY WHEN CORE INPUTS ARE PRESENT)
A. Format decision: buyer task, one next action, selected format and reason. If not a carousel, give the more suitable format outline, missing inputs and Hold/Ready rationale; stop before the card table.
B. Campaign record, using exactly these fields: company and offer; buyer and situation; one action; format decision; destination mode; exact URL(s)/form ID; allowed evidence with source/permission; claims needing approval; image assets/rights; product/claim reviewer; campaign/link/tracking reviewer; tracking plan with three separate measures, reporting window and attribution limits; first-card promise; final CTA/delivery.
C. Introductory text with character count. Aim for 150 or fewer characters. Explain any longer copy and never exceed 255.
D. Card table with exactly these columns: position; card job; exact on-card copy; visual to create or show; card headline and character count; exact destination URL or form ID; claim and source/permission; owner, review status and open question. Use 2–10 rows only. Do not pad to the user's preferred count.
E. State why card 1 works alone, trace any number promised on it to the matching middle cards, and explain the handoff from the final card to the actual destination.
F. Pre-save review: card order, image dimensions/file types/sizes/crops, mobile readability, headline and intro lengths, claims/permissions, each destination or shared form, tracking events, current Campaign Manager preview and named sign-offs. Mark each Pass, Unknown or Fail with an owner. A prompt cannot perform the preview or URL test by itself.
G. Launch decision: Ready or Hold. List every unresolved launch-critical item and owner. Say Ready only when the user has supplied evidence that the exact assets, links/events, permissions, preview and named reviews passed. Otherwise Hold.
H. After-launch observation plan: what to count for card clicks, destination/form events and qualified CRM outcomes, with reporting windows and attribution limits. Unknown remains Unknown; no invented benchmark or outcome.