URL to Markdown for Marketers: Build Swipe Files from Competitor Pages

Every senior copywriter has a swipe file — the personal vault of headlines, hooks, value props, sales letters, and launch pages worth studying. The problem with a modern swipe file is the substrate. Bookmarks rot. Screenshots clutter Drive folders no one searches. PDF prints of landing pages bury the copy under nav bars and cookie banners. Converting URLs to Markdown gives you the cleanest possible swipe file: a folder of plaintext landing pages organized by industry, funnel stage, and pattern — readable, greppable, and ready to feed to an AI for variation generation. Here is the workflow that turns a junior marketer into a competitive-intelligence operation of one.

What a swipe file is for, and why the substrate matters

A swipe file is not a copy-paste shortcut. The senior copywriters who maintain serious swipe files use them as a structural-pattern library — the lateral thinking equivalent of a chess player studying past games. You do not steal a competitor's headline; you study why their headline works, abstract the pattern, and apply the pattern to your own offer. To do that at scale, you need the source material in a format that lets you actually read, search, annotate, and compare.

The traditional substrates all have problems:

• Bookmarks: link rot. The killer landing page from 2024 is a 404 by 2026.
• Screenshots: visual fidelity at the cost of text searchability. You cannot grep an image folder for "limited-time" or "social proof".
• PDF print: preserves text but buries it in nav and footer chrome, with awkward page breaks splitting headlines from body.
• Notion clipper: better than the above, but lock-in to a single tool, and Notion's HTML-to-block converter mangles longer pages.

Markdown solves all of these. The text is clean, the structure (H1, H2, lists, blockquotes) is preserved, the file is portable across every note app and every editor, and the corpus is greppable from the command line in a single keystroke.

The swipe file structure that scales

Three top-level dimensions: industry, funnel stage, format. Most files end up indexed by all three through their folder placement and filename, with frontmatter tags carrying any cross-cutting metadata.

Swipe-File/
  By-Industry/
    SaaS/
      stripe-pricing-page-2026-05-10.md
      linear-homepage-2026-05-10.md
      notion-AI-launch-2026-05-10.md
    DTC/
      allbirds-PDP-2026-05-10.md
      glossier-launch-email-2026-05-10.md
    Info-Product/
      copyhackers-sales-page-2026-05-10.md
  By-Funnel-Stage/
    Top-of-Funnel/
      attention-grab-headlines.md
      first-touch-landing-pages/
    Mid-Funnel/
      comparison-pages/
      use-case-pages/
    Bottom-of-Funnel/
      pricing-pages/
      checkout-flows/
  By-Format/
    Long-Form-Sales-Pages/
    Email-Sequences/
    Cold-Outbound/
  Patterns/
    headline-formulas.md
    objection-handling.md
    social-proof-placements.md

The By-Industry and By-Funnel-Stage folders are where converted pages live (you may symlink the same file into both views, or pick one canonical location and tag the other). The Patterns folder is where you write your own synthesis — the abstracted insights derived from reading across the corpus.

The capture workflow

Find a landing page worth studying. Drop the URL into the URL-to-Markdown converter. The output strips the navigation bar, the footer, the cookie banner, the chat widget, and the email-capture modal — leaving the headline, sub-headline, body sections, social proof, feature lists, pricing tiers, FAQ, and CTAs. Save with a descriptive filename and the capture date.

The captured file looks like this:

---
title: Linear - The issue tracking tool you'll enjoy using
source_url: https://linear.app/
accessed: 2026-05-10
industry: SaaS
funnel_stage: top-of-funnel
format: homepage
tags: [b2b, productivity, dev-tools, hero-video]
---

# Linear is a purpose-built tool for planning and building products

Meet the new standard for modern software development.
Streamline issues, projects, and product roadmaps.

## Built for high-performance teams
...

The frontmatter carries the metadata; the body carries the copy. Both are searchable, both are AI-readable, both will outlive the live URL.

The five-pattern playbook

What you actually do with a swipe file once it has 200 entries.

Pattern 1: headline mining

Grep across the entire corpus for H1 lines, dump them into a single document, and look for structural recurrence. "The X for Y" formulas. Curiosity-gap openers. Specificity-as-credibility patterns ("3,847 marketers use..."). Reader-orientation patterns ("You want X without Y"). After you have read 100 high-converting headlines for B2B SaaS, you have an internal pattern library that produces better drafts than starting from a blank page.

Pattern 2: value-prop deconstruction

For any captured page, the second paragraph after the headline almost always reduces to: "For [audience], [product] is [category] that [primary benefit] without [common pain]." Reading 50 of these in a single session — which a Markdown corpus makes trivial — calibrates your sense of what specificity, what tone, what level of jargon works for which audience.

Pattern 3: objection mapping

FAQ sections and "common questions" blocks at the bottom of a sales page are a transcription of the prospect's actual hesitations. Aggregating FAQs across competitors in your space is competitive research. You discover the objections the market has already encountered, which informs your own page's pre-emptive answers.

Pattern 4: social-proof inventory

Where on the page is social proof placed? How is it formatted (logo wall, quote, case study card, ROI number)? What is the cumulative density (how many proof points before the first CTA)? Reading the corpus for social-proof placement teaches a structural lesson that no individual page reveals.

Pattern 5: AI-assisted variation generation

Once the corpus is in Markdown, point a frontier model at it: "Here are 30 high-converting B2B SaaS hero sections. Generate 10 hero-section variations for our product, drawing on the patterns above. Vary the angle: outcome-led, fear-led, curiosity-led." The model produces drafts that draw structural inspiration from the corpus without copying any specific page. You then iterate, edit, and run the survivors through your team.

Cross-format: competitor sales decks delivered as PDF

A meaningful share of long-form sales material is published as PDF — pitch decks, white papers, conversion guides, ebooks gated behind a form fill. If a competitor sends you their sales deck during a procurement evaluation, or if you find a public-facing PDF on their site, the same swipe-file logic applies. Convert it through PDF to Markdown and drop the output into the same folder structure. The corpus then spans both formats with no friction.

For email sequences, the workflow is slightly different — you forward the email to a tool that exports the HTML body, then convert. For paid-ad creative (Facebook Ads Library, etc.), the body copy of an ad is rarely more than a paragraph and is best captured directly as a Markdown note rather than a URL conversion.

Tooling: where Markdown plays well

The marketer-friendly note tools all read Markdown natively, which is the entire reason this substrate is durable.

• Notion: drop a folder of .md files into Notion via Import; each becomes a page. The frontmatter is converted to page properties. Tag, filter, and create database views.
• Obsidian: open the swipe-file folder as a vault. Bidirectional links between pattern notes and source pages. Graph view shows clusters. The Dataview plugin lets you query the frontmatter as a database.
• Roam, Logseq, Reflect: same story, all native Markdown.
• VS Code: for command-line natives, grep across the entire corpus is one keystroke. Multi-file find-and-replace, regex search across thousands of files instantly.

The point is that Markdown is the substrate that lets you swap tools without re-importing your library. Notion may go out of business; Obsidian may pivot; your folder of Markdown files is yours forever.

Categorization discipline

A swipe file dies the moment it becomes uncurated. The discipline is not in capturing — capturing is automatic — it is in the weekly 30-minute pass where you triage Inbox into the topical folders, write a one-sentence summary in the file's frontmatter ("Why I saved this"), and tag with the patterns you noticed.

Frontmatter tags worth standardizing across the corpus:

• format: homepage, pricing, comparison, sales-page, lead-magnet, email, ad
• angle: outcome, fear, curiosity, status, ease, contrarian
• structure: long-form, short-form, video-first, comparison-table, FAQ-driven
• proof: case-study-heavy, logo-wall, ROI-numbers, testimonial-quote, none
• cta: free-trial, demo, buy-now, lead-magnet, calculator

With these five fields populated, your corpus becomes a queryable database. "Show me every fear-angle long-form sales page in the SaaS folder with logo-wall social proof" is a Dataview query.

What to study, where to start

For a marketer building a swipe file from scratch, three sources will fill the first 100 entries fast:

1. Land-Book and Lapa.ninja: curated landing-page galleries, each entry linked to the live page. Convert each one as you browse.
2. Really Good Emails and Email Love: archived email designs from real brands. Most include the source URL.
3. Y Combinator Launch HN posts: the launch announcement of every YC company is a structured value-prop document. Reading 50 in a row teaches you the founder-to-market pitch genre.

Convert each one as you browse. After two weeks of consistent capture, you have a swipe file that beats the inherited Google Drive folder you joined the company with.

The strategic upshot

The most underrated thing a marketing team can do this quarter is build a real swipe file in a real format. The work is not glamorous — it is folder hygiene and consistent capture — but it compounds. Six months in, every new landing page brief starts with 30 minutes of corpus scanning that surfaces patterns you did not consciously remember collecting. The first draft you write is better because the inputs to your draft are better.

For mixed corpora — URLs and PDFs — see also URL to Markdown for academic research, which describes a parallel discipline applied to research sources, and URL to Markdown for legal evidence, which applies the same capture mindset to litigation work. The substrate is the same; the use case differs.