How to build a swipe file: what to save, how to tag and study it, and where the legal line is

A swipe file is a collection of other people's advertising that you save to study: sales letters, ads, VSL leads, offers and presells. To build one that pays off, save only pieces you can explain, tag each one so you can find it again, study the best by copying them out by hand, and take ideas and structure from them, never the words, claims or testimonials.

Inc., quoting Buffer in a 2014 article, described swipe files as starting with copywriters who kept advertising letters that had been tested and proven, and suggested an email folder or label with sub-folders for whatever inspired each piece (Inc.). A folder is only storage, though. The method has five stages:

From saved to tested

The five stages of a swipe file that teaches you something

1. 1Collect
2. 2Tag
3. 3Study
4. 4Apply
5. 5Test

1Save pieces you can explain

What happens: VSL leads, mechanisms, offers, ads and presells that seem to be working for someone, or whose effect you can put into words.

Why it matters: A swipe you can't explain can't teach you anything.

What goes wrong:

• Saving everything you like
• Saving only from your own market

2Tag it the day you save it

What happens: Awareness level, lead type, hook, mechanism, offer, format and source, plus one line on why it works.

Why it matters: Tags are what make a few hundred entries searchable.

What goes wrong:

• Tagging later, which rarely happens
• Inventing new labels every week

3Copy the best out, then time it

What happens: Write the lead out word for word, then note each choice the writer made and the second it lands.

Why it matters: Copying slows you down enough to see the choices the writer made.

4Write the principle into your own script

What happens: Name the move in one sentence, then write it fresh for your product in your buyers' words.

What goes wrong:

• Changing a few words of someone else's lead
• Borrowing their claims along with their structure

5Treat it as a hypothesis

What happens: Run the new lead or offer against your current one on the same traffic.

Why it matters: It worked for someone else's product and audience. Your traffic decides whether it works for yours.

Step 1: Collect pieces you can explain

Save what teaches you something. For VSL copy, that means five kinds of material:

• VSL leads. The headline, the poster and the first 60 to 90 seconds, transcribed. That's where viewers decide whether to stay.
• Mechanisms. How a product is explained to work, and what the explanation is called.
• Offers. Bundles, price per unit, guarantees, bonuses, subscription terms, and the moment the price appears.
• Ads. The hook, the angle, the format and the page each ad sends people to.
• Presells. The article or page between the ad and the VSL, and the job it does there.

Leave out what you merely like. When you're unsure about a piece, three questions settle it:

Save it or not?

1Can you say in one sentence why this piece works?

YesNo, I just like it

Not that you like it: what it does that others don't.

2Is there any sign it's working for its advertiser?

Some signNone

An ad still running months after its start date, or a VSL you keep meeting from new ads. Clues, not proof.

3Does its pull depend on a claim, a testimonial or a result you couldn't prove for your own product?

YesNo

Best fit

Skip it for now

You like it but can't say why it works, so it can't teach you anything yet. Come back to it when you can name the move it makes.

Best fit

Save it, tagged as untested

You can say why it works, but nothing suggests it's working for its advertiser. File it as an idea rather than a proven swipe, so you don't treat it as evidence later.

Best fit

Save the structure, mark the claims as theirs

You can say why it works and something suggests it's working, but its pull rests on claims, testimonials or results you couldn't prove. Keep the shape of the argument and flag the claims as off-limits.

Best fit

Save and tag it in full

You can say why it works, something suggests it's working, and its pull doesn't depend on proof you lack. Tag it today, while you still remember why you saved it.

Where to find ads and VSLs worth saving

Start with Meta's Ad Library. Since a 2019 change it shows every active ad that any Page is running, not only political ones, and you can search by advertiser as well as by keyword (Meta, 2019). Three facts about the library change what you record:

• It shows no results. For an ordinary commercial ad there is no spend, no impressions and no performance metric at all; spend and impressions appear, in ranges, only for ads about social issues, elections or politics (Meta's developer reference).
• The start date is a plan. In Meta's documentation, an ad's start date is when the advertiser wanted it to start delivering. A long run is a clue that an ad may be paying its way, not proof, so record the start date and the last day you saw the ad running, and treat the rest as a guess.
• Keywords aren't translated. Meta's reference for searching the library says it doesn't translate keyword searches, so search in the language of the ads you want (Meta's developer reference).

Ads shown in the EU carry more detail. Meta said in 2023 that ads targeting people in the EU would stay in the library for a year, with the dates they ran, the targeting parameters used and who they were shown to (Meta, 2023). Record the targeting for those too. And if you advertise in the EU yourself, competitors can read yours.

Beyond the library, click through to the pages the ads send people to and save each VSL's opening. Keep competitors' order pages. Above all, keep your own past tests, the one kind of swipe whose numbers you actually know.

Step 2: Tag every entry the same way

Tags turn a pile into a swipe file. Use the same fields for every entry, in whatever tool you'll actually search; a spreadsheet with one row per piece is enough. The fields that matter for VSL copy:

• Awareness level: unaware, problem aware, solution aware, product aware or most aware.
• Lead type: one of the six named in Great Leads (offer, promise, problem-solution, secret, proclamation or story). A fixed list is what makes the column searchable.
• Hook: the move the first line makes, named from one fixed list, such as the 24 formulas in our VSL hook examples.
• Mechanism: the reason given for why the product works, in one line.
• Offer: the bundle and price structure, the guarantee, bonuses and any subscription.
• Format and market: presenter, slides or UGC style; the length; the category and price point.
• Source and dates: where you found it, the start date shown and the last day you saw it running.
• Why it works: one sentence the day you save it, and the details once you've studied it.
• Hands off: the claims, testimonials and lines that stay theirs.

One entry with every field filled in, for a fictional product:

One swipe file entry, tagged

A fictional gutter-guard VSL, saved from the Ad Library

The swipeTagsWhy it worksHands offYour test

What was saved

Page headline: "If you drag the ladder out every November, watch this before you climb it again."
First spoken line: "Last fall I watched my neighbor wobble on his ladder, scooping wet leaves out of his gutters. Here's the five-minute fix that means I never go up there."

Also saved: the poster, a transcript of the first 90 seconds, the offer section and the landing page.

Tags

• Awareness: problem aware
• Lead type: story
• Hook: an "If you" callout on the page, then a story in progress
• Mechanism: a fine mesh that sheds leaves instead of catching them
• Offer: kits sized by gutter length, a lifetime warranty, free shipping on two kits
• Format and market: presenter VSL, 14 minutes; home improvement, orders of $150 to $400
• Source and dates: Meta Ad Library; start date shown 12 November; saved 3 March, still running

Why it works

It opens on a moment every homeowner with gutters knows, the ladder coming out, before it mentions any product. The story belongs to the presenter, so the mechanism arrives as something he found rather than something he sells. And the offer is sized by the buyer's own house, which turns the price into a measurement.

Hands off

• Their words: the headline and the script are their expression.
• Their story: the neighbor and the ladder are theirs to tell.
• Their claims: "never clean your gutters again" rests on their evidence, and you'd need your own.
• Their testimonials and their footage, under any name.

Your test

Hypothesis: a problem-aware story lead about the ladder moment, told by our founder about her own house, will hold more viewers to 60 seconds than our current lead, which opens on the product. Test it as a new opening on the same body.

Write the why-it-works line on the day. A month later you won't remember what caught your eye, and an entry without it is just another screenshot.

Step 3: Study the best by copying them out

One study method has a clear source, and it's slow on purpose. Gary Halbert asked readers of his newsletter to go further than reading classic sales pitches: copy them out word for word in their own handwriting, over and over, until good writing became automatic. He was explicit about the purpose: the idea was "not for you to steal these promotions but rather, to be influenced by them" (The Gary Halbert Letter). One of the pieces that issue reprinted for copying was David Ogilvy's Rolls-Royce ad (what Ogilvy still teaches). In an update to another issue, his son Bond adds that hand-copying isn't required to become a great copywriter, though for many people it gets them there faster (The Gary Halbert Letter). Our guide to the Boron Letters covers his other lessons.

The point of the slowness is attention. You can't skim while you copy, so you notice the choices: where a sentence breaks, which word opens a paragraph, how long the writer waits before the product appears.

For a VSL, transcribe the lead, then copy the transcript out by hand. Afterwards, write the choices you noticed under why it works, each with the time it lands in the video: the product named at 1:40, the first proof at 2:15. Those times are what you'll compare across entries; our VSL examples guide turns them into a full beat map.

Step 4: Apply the principle, not the words

A swipe earns its keep once you stop looking at it. Four moves:

1. Name the principle in one sentence that uses none of the original's words, such as: open on the moment the problem shows up, before naming a product.
2. Close the file.
3. Write your version from your own research, in your buyers' words, with your own presenter's true story.
4. Check each claim against your own evidence, then fit the lead into the beats of your script (how to write a VSL script).

The difference shows in a single line. A copy changes only the nouns: "If you drag the snow shovel out every December, watch this first." Applying the principle keeps the move and nothing else: a lawn-care brand might open on the Saturday its owner spends pulling weeds by hand, in the words its own customers use.

What copyright lets you take from a swipe file

The line is drawn in the statute. Under 17 U.S.C. § 102(b), copyright protection never extends to an idea, procedure, process, system, method of operation, concept, principle or discovery, however it's described (17 U.S. Code § 102). The US Copyright Office's Circular 33 says it in plain English: copyright protects the original expression in a work, not the idea underneath, and words and short phrases such as names, titles and slogans can't be copyrighted (Circular 33). Copyright isn't the only rule that applies, though:

What a swipe file lets you take

Ideas, not expressionYours to use

• The structure: problem, then story, then mechanism
• The lead type and the angle
• The idea of a mechanism, if it is true of your product and explained your own way
• An offer structure: bundles, a guarantee, a bonus
• Methods: when the price appears, where the proof goes

Expression, and moreNot yours

• The words of the script, the headline and the ad
• The footage, the images and the page itself
• Their claims, which need your own evidence
• Their testimonials and reviews
• Their brand names and taglines, which can be trademarks

Copyright covers the first two. Advertising law and trademarks cover the rest.

Two of those limits come from the FTC rather than copyright. Its rule on consumer reviews and testimonials, in force since 21 October 2024, prohibits fake or false consumer reviews and testimonials (Federal Register), and a testimonial lifted from another brand's ad and shown as your customer's misrepresents whose experience it is. A claim isn't evidence either: for health claims, the FTC expects competent and reliable scientific evidence that matches the product being sold (FTC).

Old ads are a different case. As of 1 January 2026, works published in the US in 1930 or earlier are in the US public domain, under the 95-year term for works published before 1978 (Duke Law's Center for the Study of the Public Domain). By that rule, Claude Hopkins' Scientific Advertising (1923) is free to copy in the US. Anything published from 1931 to 1963, such as Caples' Tested Advertising Methods (1932) or Ogilvy's Confessions (1963), is still protected if its copyright was renewed; later books, Schwartz's Breakthrough Advertising and Ogilvy on Advertising among them, are protected. New editions carry their own copyright in new material, and other countries set other terms. This isn't legal advice.

Step 5: Test what you borrowed

A swipe shows what worked for someone else, with their product, their audience and their offer, if it worked at all. For you it's a hypothesis. Write the test before you write the lead: what you expect to change, the number that will show it (viewers still watching at one minute for a lead, revenue per visitor for an offer) and how much traffic you'll need to tell.

In RevvyScales, a folder splits new visitors between whole VSL versions by the weight you set, and each visitor keeps seeing the same one, so a lead you built from a swipe can run against your current lead on the same traffic (folders and A/B splitting). When the result is in, write it in the swipe file next to the entry that inspired it. A swipe with your own numbers beside it is worth ten without.

Set up your swipe file this week

Eight tasks, in order; the ticks stay in this browser:

Your swipe file, set up in a week

• Create one sheet with the fields above, from awareness level to hands off.
• Save ten pieces from your market and two from a market next to it.
• Tag all of them the same day, with a why-it-works line for each.
• Search Meta's Ad Library in your buyers' language, and record each ad's start date as a clue, not proof.
• Copy one strong lead out by hand, start to finish, and note when each choice lands.
• Write one new lead for your product from a principle you found, with the file closed.
• Mark every claim and testimonial in your swipes as theirs.
• Put the new lead in a test, and log the result next to the entry that inspired it.

0 of 8 done