Messaging teardown template

The template

Copy this block into a doc. It is the output format of our own teardown method, the one the Teardown tool on this site runs.

# Teardown: <Company>, <page type>

**Source:** <full public URL>
**Fetched:** YYYY-MM-DD
**Human ingredient:** my opinion on why this page fails, given YYYY-MM-DD

> <your opinion, verbatim, one to three sentences>

## Before
The page as it read on <fetch date>. Quoted, not paraphrased.
- **Hero headline:** "<verbatim>"
- **Subhead:** "<verbatim>"
- **Primary CTA:** "<verbatim label>"
- **Secondary CTA:** "<verbatim label, or none>"
- **Section headers, in order:** <verbatim list>
- **Proof on the page:** <logos, numbers, named customers, quotes, or none>

## The Eight Reads
| Read | What the page does | Evidence, quoted verbatim | Mark |
|---|---|---|---|
| 1. The Reader | | "" | Holds / Soft / Fails |
| 2. The Category | | "" | |
| 3. The Claim | <the claim, and its negation> | "" | |
| 4. The Ask | | "" | |
| 5. The Clock | | "" | |
| 6. The Vocabulary | | "" | |
| 7. The Proof | | "" | |
| 8. The Swap | | see below | |

**Read 8 in full, the swap test.**
- "<line>" could run unchanged on <Competitor A>.
- "<line>" could run unchanged on <Competitor B>.
- Lines that could not: "<line>", because <reason>.

## Diagnosis
**Primary failure: Read <N>, <one precise sentence>.**
<the quoted copy that proves it>
**What it costs them.** <which visitor, from where, leaves at which moment, and does what instead>
**Secondary.** 1. <one line> 2. <one line>
**The fair reading.** <the best reason a competent team shipped this, and why the diagnosis survives it>

## After
Replacement copy, not advice. For each of hero headline, subhead,
primary CTA and the one section that fixes the primary failure:
Before: "<verbatim>"  After: "<replacement>"  Why: <one line, tied to a Read>

## Does the rewrite pass the swap test?
<Run Read 8 on the new copy. Name the competitor that could not publish it, and why.>

## What a reader can steal from this
- <two to four bullets: the method, not the conclusion>

## Sources
- <page URL>, fetched YYYY-MM-DD
- <Competitor A URL>, fetched YYYY-MM-DD
- <Competitor B URL>, fetched YYYY-MM-DD
- <public source URL for any general marketing claim made above>

How to fill each slot

The output has these parts, in this order. Each slot names the rule that decides when it is done.

1. Header. Write your opinion before you fetch anything, quote it as you wrote it, and date it. Without it the method writes nothing and returns one line: "Teardown blocked. Need your opinion on why <URL> fails. One or two sentences is enough."
2. Before. Copy the hero, every button label in order, the section headers and the proof, word for word. Keep their typos and their em dashes. Then check each quote against the raw HTML. If the two disagree, the markup wins.
3. The Eight Reads. Give Reads 1 to 7 a mark and at least one verbatim quote each. The method never calls a mark a verdict: a mark is evidence, and the diagnosis is the verdict. Leave no row blank, and keep per-Read notes inside the table.
4. Read 8. Score it only after you fetch two named competitors and date both pages. Quote each line a rival could run, and name the rival.
5. Diagnosis. Name one primary failure, picked by your opinion, not by the harshest mark. The cost sentence names one visitor and the moment they leave. Add at most two secondaries and the fair reading.
6. After. Write the replacement lines themselves, each beside the line it replaces. Every new claim must be falsifiable and supported by the company's own public site.
7. The swap test on the rewrite. Run Read 8 again on your new copy, and name the rival that could not publish it.
8. What a reader can steal. Write two to four bullets of method. The method's test: delete the company name, and this section must still stand on its own.
9. Sources. List every URL with its fetch date, for the page and both rivals. Any general claim about marketing practice gets a public source URL here too.

The blocked version. Two gates can stop a run after the Reads are scored. One fires when the target's own site announces a merger or acquisition. The other fires when your opinion maps to no Read that failed. End the file name in -blocked.md and open it with a blocking header. It names the gate, asks the one question you need answered, and says the file is not ready for publication. If you already gave your opinion, quote it under the header. Keep the header, Before, the Eight Reads and Sources. Omit Diagnosis, After, the rewrite's swap test and What a reader can steal entirely. The method's reason: an omitted section is clear, and an empty one invites someone to fill it.

Worked sample: Figma's homepage, filled in

"The full-stack creative canvas"

Button: "Get started"

"Move fast in the right direction on an AI-native canvas."

"Teammates and AI agents work in the same space with shared context. Generate new ideas, refine, and align—together."

Read on 2026-10-14. Source: Figma homepage

That is the H1, its button, and the first section under it. Further down, the proof reads:

"95% of the Fortune 500 uses Figma"

"Based on data from March 2025."

Read on 2026-10-14. Source: Figma homepage

The browser tab title, which is not hero copy, reads "Figma: The collaborative canvas for design, code, and AI". We checked every quote against the raw HTML.

What the copy does well. The proof holds. It is a number with a unit and a date, and it names the brand. A named customer quote sits beside it, from Francisco Seiz, "Senior Design Director, Code and Theory". Inside the hero, the page asks for one thing, "Get started", and asks for little.

The template, filled in. We wrote the opinion first. The rivals are Penpot and Sketch, both read on 2026-10-14.

# Teardown: Figma, homepage
Source: https://www.figma.com/   Fetched: 2026-10-14
Human ingredient: our opinion, given 2026-10-14
> "Penpot could put its logo on this headline tomorrow. The hero says
>  what Figma is, a canvas, but not what only Figma does."

## The Eight Reads
| 1. Reader     | "Teammates and AI agents": no role named         | Fails |
| 2. Category   | "canvas"; the tab adds "for design, code, and AI" | Soft  |
| 3. Claim      | "Move fast in the right direction"; its negation | Fails |
|               | "move slowly in the wrong direction" is a line  |       |
|               | no company would publish                         |       |
| 4. Ask        | one low-cost ask in the hero, "Get started";    | Holds |
|               | "Contact sales" sits in the nav, beside         |       |
|               | another "Get started"                            |       |
| 5. Clock      | no reason to act this quarter                    | Fails |
| 6. Vocabulary | "Explore Figma Make" sits under a line naming    | Soft  |
|               | the job, but the button carries the product     |       |
|               | name, not the job                                |       |
| 7. Proof      | "95% of the Fortune 500 uses Figma", dated       | Holds |
| 8. Swap       | see below                                        | Fails |

Read 8:
- "The full-stack creative canvas" could run unchanged on Penpot, whose own
  section reads "Full-stack design for every future."
- "Powerfully expressive. Incredibly precise." could run unchanged on Sketch,
  whose menu reads "Create with precision".
- Could not: "95% of the Fortune 500 uses Figma", because it names
  Figma and dates its own figure.

## Diagnosis
Primary failure: Read 8. The H1 claims its category with the word
a named rival already uses: "full-stack".
What it costs them (our sketch): a design lead comparing Figma and
Penpot reads "full-stack" on both pages and leaves for the pricing
pages to decide on cost.
Secondary: 1. Read 1, no role named. 2. Read 3, the claim cannot
be proven wrong.
The fair reading: Figma sells design and build tools ("Explore
design tools", "Explore build tools"), and a broad line covers both.
It still fails the swap, because Penpot claims the same breadth:
"BUILT FOR DESIGNERS, DEVELOPERS AND AI AGENTS".

## After
Headline. Before: "The full-stack creative canvas"
  After: "The design canvas 95% of the Fortune 500 uses."
Subhead. Before: "Move fast in the right direction on an AI-native canvas."
  After: "Designers, developers and Figma's agent work in one file,
  from the first idea to the code that ships."
CTA. Keep "Get started". The Ask held.

## Does the rewrite pass the swap test?
Penpot and Sketch could not publish the headline: it names Figma's
own figure, dated March 2025 on the page. The subhead could move to
Penpot with one name changed, so the headline carries the difference.

We left the fourth After slot, the one section that fixes the primary failure, out of this short version. Before sits in the quotes above it, the steal section follows it, and Sources are at the end of this page. In our view, the fix is the headline itself, because it moves the page's own proof to the top. Read 3 tests the hero's reason to choose; the 95% figure is proof, so Read 7 scores it. Every claim in the rewrite comes from Figma's page: the figure, "Ask Figma’s agent", "Dev tools that take you all the way to production" and "without ever leaving the design file". The cost sentence is our sketch, not a claim about who visits Figma.

What a reader can steal. Swap test your H1 against the rival a buyer opens next. If your strongest proof sits below the fold, try it as the headline.

Common mistakes

• Filling the header last. The method treats the opinion as the input that picks the primary failure. Write and date it before you fetch the page.
• Writing "verdict" in the Mark column. The method uses only Holds, Soft or Fails there. The diagnosis is where the ruling goes.
• Leaving blocked sections empty. A blocked teardown omits the four sections outright. Keep the fetch dates, because that is the part that goes stale.
• A steal section about one company. If the bullets stop making sense once you delete the company name, the method calls the piece an opinion about one company. Rewrite them as method.
• Advice in the After. "Lead with proof" is a note. The method asks for the replacement line itself.

Questions people ask

Is there a template for a homepage teardown?

Yes. The block at the top of this page is the format our method uses on any public B2B homepage. It runs from a dated header to the Eight Reads, a diagnosis, the after and sources. The method covers positioning and pricing pages too. In our view, the slot to check in any template is the dated opinion in the header.

Is there a messaging teardown template in PDF or Word?

The block above is plain markdown text. Paste it into the doc tool you use, check that the Eight Reads table came through, and export a PDF from there. In our view, an editable file beats a PDF here, because each teardown is re-run as a new dated file when the page changes.

Sources

• Figma homepage, https://www.figma.com/, read 2026-10-14
• Penpot homepage, https://penpot.app/, read 2026-10-14
• Sketch homepage, https://www.sketch.com/, read 2026-10-14
• Google results for "messaging teardown template" (People also search for: "Messaging teardown template pdf", "Messaging teardown template word"), https://www.google.com/search?q=messaging+teardown+template, read 2026-10-14
• Our teardown method skill and its output format, as used by the Teardown tool (ours, read 2026-10-14)