You are writing an original LinkedIn Document Ad for a B2B company. The inspected reference below is a structural example, not permission to copy its words, branding, statistics or conclusions. Work from the user inputs and label every assumption.

REFERENCE INSPECTED
KPMG native Document Ad, LinkedIn Ad Library 1583754176, captured 2026-09-23. Its cover asks who generates higher technology ROI and promises four patterns from a Global Tech Report. Preview page 2 centers a 3.6x figure about smaller organizations with a short explanation; page 3 compares 2.2x and 1.4x for early and late adopters. Three archive pages were saved. The live ad viewer was also inspected through page 6: page 4 shows 2.6x ROI for organizations facing fewer cost pressures; page 5 shows 3.2x ROI for transformation-focused organizations and says 50% or more of their tech budgets go to transformation; page 6 directs readers to the full Global Tech Report 2026. These are KPMG report claims. Gate and ad results are unknown. Borrow the one-finding-per-page structure, not the figures.

USER INPUTS TO COLLECT OR USE
Company and offer: [ ]
Specific buyer and buying situation: [ ]
One document promise and desired next action: [ ]
Source material with URLs/files, dates and publication permission: [ ]
Approved customer names/quotes/figures, if relevant: [ ]
Available format: source report, checklist, case, survey or guide: [ ]
Free reading or Lead Gen Form, and why: [ ]
Asset/thank-you URL, privacy URL, CRM owner and follow-up sender: [ ]
Brand constraints and preferred page count: [ ]

SPECIFIC ASSIGNMENT
Make a research-led document in which each page states one verified finding, its comparison and what the reader can do with it. Preserve sample, date, denominator and source. Do not infer cause from correlation. If there are fewer than three findings, build around the available findings instead of inventing more.

WORKED PATTERN, CLEARLY HYPOTHETICAL
Hypothetical inventory-planning company: cover “Where do replenishment plans break? Three patterns from our 2026 inventory review.” Page 2 shows a supplier-update timeline instead of a fabricated percentage. Page 3 shows an exception owner checking changes before purchase orders run. A number is added only after the underlying analysis is supplied.

MISSING-INPUT RULE
If material is missing, ask up to three focused questions, then continue with a clearly labeled draft using bracketed placeholders. Do not invent a business fact, customer, quote, survey result, performance rate, source, permission or link. If the core proof is absent, give a non-numeric/process alternative and mark evidence-dependent pages “not ready to publish.” Do not claim the inspected advertiser used a gate or achieved a result.

OUTPUT CONTRACT
1. State the buyer, one-sentence document promise, why this format fits, and the proposed free/gated choice. Separate what is known from assumptions.
2. Give a complete 6–8-page document. For EVERY page provide a page number, exact headline and body copy ready to paste, what the reader learns, layout instructions that work at phone width, and the source/approval needed for each factual claim. Pages 2–3 must fulfill part of the cover promise, not just introduce the company. Mark the proposed preview boundary and show what remains after it. Do not fill pages with slogans or generic benefits.
3. Write the sponsored intro (aim under 150 characters where possible), a concise headline, and a plain-text end CTA. Make their promises match the document.
4. If gated, specify minimal form fields, the privacy URL placeholder, the exact confirmation message, full-asset URL placeholder, CRM routing owner and a test submission. If free, explain how a reader gets the PDF and what next action is offered. Do not assert that PDF links work inside the LinkedIn feed; print essential URLs visibly and test after download.
5. Write one delivery email or follow-up message that refers to a specific page. Distinguish an educational download from a request to speak with sales.
6. Finish with an evidence/permission checklist, a PDF-export and mobile-readability checklist, and three things to inspect after launch: page consumption, successful delivery, and qualified next steps. No invented benchmarks.
Return actual copy and page layouts, not advice about what someone could write.