How to critique my homepage messaging
To critique your own homepage messaging, start with a one-line opinion on why it fails, ideally from someone else. Then read the live page eight ways: reader, category, claim, ask, clock, vocabulary, proof and swap. Name one primary failure from that opinion, and rewrite only the lines it condemns, using proof already public on your site.
Last updated by Omair Dawood.
The eight changes when the page is yours
The Teardown method is ours. It is the nine-step process behind the Teardown tool, and the full version is in How to do a messaging teardown. The method was written for public commentary on someone else's page. The first change below is its scope rule. The other seven are method steps that work differently when you own the page, in the method's order.
- Decide which job you are doing. The method's scope rule says a teardown is public commentary on a page you do not own. Replacement copy for your own page, or a client's, is a different job: commissioned copy. On your own homepage, the diagnosis is the deliverable. Treat the rewritten lines as a brief for whoever writes the final copy.
- Get the opinion from outside the room. The method will not run without a person's one- or two-sentence opinion on why the page fails. In our view, an outside opinion sharpens the diagnosis on your own page, because the writer reads the intent into each line. Ask a colleague in sales or a customer to write the sentence before you read the page again.
- Read the live page, not the draft. The method quotes the page word for word and records the date read. The Teardown tool only reads public addresses. For localhost, a bare IP address or a
.localhost, it replies "The tool reads public websites only." So the critique stays on the words a visitor sees. - Mark each Read from the words alone. Each of the Eight Reads gets Holds, Soft or Fails, with a quote as evidence. Read 1 asks who the copy assumes is reading "from the words on the page alone." Mark what the sentence says, not what you meant it to say.
- Fetch two named competitors for the swap test. Read 8 puts a competitor's logo on your page and checks which lines survive. The method fetches both competitor pages and records their dates, because a swap test from memory is a guess. That holds even when you know your rivals' copy well.
- Let the opinion pick the primary failure. The method picks it from the opinion you give, not from the severity of the marks. If two Reads fail, the one the opinion points at becomes primary, and the other becomes a secondary. So the person whose sentence you feed in steers the whole result.
- Rewrite only from public proof. Every replacement claim must rest on evidence already public on your own site. Nothing invented, no new numbers. A win that is not on your site yet cannot go in the rewrite. In our view, when your strongest proof is missing from the page, publishing it is the first fix.
- Steelman with the constraint you lived through. The fairness pass asks for the best reason a competent team shipped the page. If you know the real constraint behind a line, write it down as that reason, then check the diagnosis still stands.
Worked sample: Gong's homepage
"#1 AI OS for Revenue Teams"
"Revenue AI Built To Predict pipeline risk"
"Capture every interaction, analyze what's working, and automate what happens next — ring the gong, deal after deal, quarter after quarter."
Read on . Source: Gong homepage
The headline's last words rotate. The page markup lists six endings: "Close deals faster, Predict pipeline risk, Automate handoffs, Enable at scale, Increase win rates, Predict churn". The quote above shows the ending on screen in our capture. The rotation means a visitor may see any of the six. Further down, a section headed "Everyone wins with Gong" says:
"From execs to RevOps to frontline reps, bring your entire GTM team together in Gong to move faster, work smarter, and unlock sustained growth."
Read on . Source: Gong homepage
What the copy does well. The page commits to a category. The browser title reads "Gong - Revenue AI OS", and the eyebrow line names "AI OS", so Read 2 holds. Read 7 holds too. The customer quotes carry names and job titles, such as a VP of Sales Enablement at ADP. The subhead names three plain actions: capture, analyze, automate.
The opinion, and where it points. Here is ours, given 2026-10-13: "The headline rotates through six jobs. The job a reader sees depends on when they look, and the page never says which reader comes first." It maps to Read 1, the Reader. In our view, the second quote confirms it: "execs", "RevOps" and "frontline reps" are three seats who come to the page for different things. The hero also carries two asks, "Book a demo" and "See Gong in action", which could mark Read 4 as soft. Under step 6, Read 1 is still primary, because the opinion points there. Read 4 becomes a secondary.
The fix. An outsider can rewrite the words but cannot choose the reader. So the headline keeps one slot open.
Before: "Revenue AI Built To" plus six rotating endings. After: "Revenue AI built to [the one job your first reader came for: an owner's decision]"
Before: "Capture every interaction, analyze what's working, and automate what happens next — ring the gong, deal after deal, quarter after quarter." After: "Gong captures every customer interaction, shows what's working and automates the next step."
The subhead rewrite uses only the page's own claims. The open slot is the point of this guide. An outside critic stops there, because naming Gong's first reader would be a guess. The page's owner can fill it.
Common mistakes
- Marking intent instead of words. In our view, the writer of a vague line tends to read the intended buyer into it. Score the sentence as a stranger reads it.
- Writing the opinion after the Reads. In our view, an opinion written after the marks bends to match the worst one. Write the sentence first and keep it in view.
- Feeding in a soft opinion. "It could be clearer" maps to no Read. The method escalates an opinion that matches no failing Read and asks which one you meant.
- Rewriting every line. The method rewrites only what the diagnosis condemned, so the diagnosis and the fix read as one argument.
- Citing proof the page does not show. A replacement claim needs evidence already on your site. Publish the case study, then write the line.
- Skipping the competitor fetch. Read 8 needs two real pages, quoted and dated. The method treats a swap from memory as a guess.
Questions people ask
Can someone critique my homepage?
Yes. Anyone who did not write the page can give the opinion the method needs, in one or two sentences. The Teardown tool reads any public URL, your own homepage included. It returns the Eight Reads, one primary failure and replacement copy. The method's scope rule treats commissioned copy as a different job. In our view, use that replacement copy as a brief for the final version.
How do you critique a value proposition headline?
Use Read 3, the Claim. Pull the one sentence meant to be the reason to choose you, then write its negation. If no company would ever say the negation, the headline is decoration. Then run Read 8: put two named competitors' logos on the page and see if the line survives. A headline that passes both is falsifiable and yours alone.
Sources
- Gong homepage, https://www.gong.io/, read 2026-10-13
- Web search results for "how to critique a homepage value proposition headline", read 2026-10-13
- Our Teardown method skill and the Teardown tool on this site (ours, read 2026-10-14)
Critique your homepage with the Teardown tool
The tool asks three things: the page URL, why it fails in a sentence or two, and two competitors. Leave the competitors blank and it finds two. You get four things. The page's copy is quoted as it reads today, and each of the Eight Reads is marked Holds, Soft or Fails. Then come one primary failure with what it costs, and replacement copy line by line. It takes about 2 minutes of typing.