How to test a headline change without changing the offer
A headline test keeps the buyer, the promise, and the button still, and changes the sentence. Here is how to submit that pair and how to read comments that wander off the words.
Picture Ruth, a writer in Dublin, duplicating a homepage file on a Tuesday and changing one noun. The offer stays a collections inbox for finance leads. The button stays "See a sample inbox." The new headline names the Friday close instead of naming the inbox. She is about to "tidy" the illustration because the new line is shorter, and to soften the button because the new line feels warmer. Those tidies would end the test. She would learn whether people crush a different page, not whether they crush a different sentence.
A headline test is the smallest variant worth submitting. LandingCrush can show each sentence as its own approved homepage and report a crush rate, a count, and comments. It cannot know that you meant to change only the words if the picture moved. You protect the meaning by freezing the offer. The general rule is one difference. This article is that rule applied to the sentence at the top.
Write the offer on a line the headline is not allowed to break
Before you draft B, write the offer underneath both files. Who it is for. What they get. What they are asked to do. The headline may emphasize any part of that line. It may not recruit a new buyer, invent a new outcome, or rename the action. "See a sample inbox" cannot become "Book a demo" inside a headline test. That is an offer test wearing a writer's badge. Offer tests are legitimate. They need their own sentence and their own readout. Do not mix them into a copy tweak.
Read both headlines against the offer line word by word. Circle any noun that is not in the offer. A circled noun is either a clarification of the same buyer or a smuggled pivot. "Controllers at 50-person firms" clarifies finance leads if that is truly who you serve. "Startup founders" abandons them. If you want to know whether founders prefer a different story, say so. Do not call it a polish.
Keep the claim strength still. If A does not promise a time saved, B should not promise three hours. A stronger claim is a different offer even when the product is unchanged. Voters may crush the bigger promise. You would learn that people prefer bigger promises, which you already knew, and you would not learn whether the sentence was clearer.
Freeze everything the eye meets except the words
Same logo, same picture, same button label, same proof, same navigation, same background. Line length is the trap. A shorter headline reflows the hero and moves the button. Rebuild the frame so the button stays where it was. If you cannot, write the leak into the test sentence: "B also moves the button up because the line is one row shorter." Then decide if you still want to run it. A known leak is better than a pretend pure test. An unknown leak shows up later as comments about the layout when you wanted comments about the noun.
Do this on the mobile capture too. A headline that fits on a desktop and truncates on a phone is not the same sentence. Check the phone file before you submit. Voters who meet the phone capture are voting on the words they can see. If the last half of the buyer disappears, you tested a fragment.
Run the five-second check from how to score a SaaS hero on both, logo covered. You should be able to say the same offer from both screens, with a different emphasis. If B makes you describe a different product, the headline broke the line. Fix B. Do not submit a broken line because the draft is clever.
You might be thinking headlines have to change the feeling or they are pointless. Feeling can change while the offer stays. "The Friday close, without the spreadsheet hunt" feels different from "Collections inbox for finance leads" and can still describe the same product and the same button. The test is whether voters crush that emphasis. It is not whether you personally find it warmer.
Submit both sentences as their own homepages
Upload or capture each file at /submit. Name them so the dashboard shows the emphasis. "Headline: inbox" and "Headline: Friday close" will still make sense in three weeks. "New new final" will not. Wait for both approvals. The free plan includes one active submission. Holding two headline variants live is a Creator Pro situation, at $19 a month or $190 a year, because that plan includes unlimited submissions. If you overwrite the only slot, you are not running a paired test.
Decide the window before the rates move. Either let both collect in the feed until each has a count you respect, or start a 48-hour duel if you need a shared link. The choice between those instruments is in duel or two live submissions. Do not switch stories halfway. A duel link and an All-Time rate are different samples.
Ask people who already love the brand to use Blind Mode if you want the words to carry the vote. The logo is identical on both pages, so recognition helps both. It still flatters a weak sentence when the voter is your friend. The blind setup is in how to use Blind Mode.
Read comments for the noun you changed
When you open the detail pages, search the comments for the words you actually changed. If you emphasized the Friday close, you want to see whether people repeat that idea or ignore it. A crush rate that moves while every comment talks about the illustration means the leak you hoped was small was the whole vote. Report the leak. Do not write "the Friday headline won."
If comments say both headlines are the same, believe them. You may have changed a word that does not change the meaning for a stranger. That is a useful negative. It means the next test has to be a larger difference in the sentence, still inside the offer, or you should stop headline testing and work on proof. More adjectives will not create a difference voters can see.
If the rates are close and the comments split, the outcome is no preference. Keep the sentence that is more accurate, or the one support can say on a call without wincing. Accuracy is a fine tie-breaker. Cleverness is a weak one. A close rate is not a failure of the method. The method's job was to stop you from shipping a reflow and calling it copy.
Compare counts before you compare percentages. A headline "win" of eight votes against seventy is not a win. Wait, or say the sample is directional. Fifty votes is the neighborhood LandingCrush uses before it treats a community result as ready to audit against a panel prediction. You can act earlier. You cannot act loudly.
Leave the AI Panel out of the headline readout unless you label it a preview. A model that prefers the longer sentence has not voted. If you bought the $29 or $49 test, park it in the other column described in the panel article. The sentence you ship should be one you can point at in the comments or one you kept because there was no preference.
Ship the sentence, then watch the offer you refused to change
Put the winning sentence, or the kept sentence, on the live homepage. Do not quietly change the button in the same release. You would bury the only thing you learned. Watch the action named in the offer line, in your analytics, for a period you choose in advance. LandingCrush will not show you that action. It showed you which submitted sentence voters crushed, if the counts allowed the claim.
If the live metric moves the wrong way, you learned something the swipe could not know: the voters in the feed and the visitors on your site are not the same people. Keep the note. The next headline test should not pretend otherwise. Write the offer line again. Change one sentence. Leave the rest of the hero where it was.
Which noun are you actually testing, and which noun did you promise not to touch?
Common questions
Q: Can the button label change in a headline test?
A: Not if you want to learn about the headline. A new button is a new action. Freeze it, or rename the test.
Q: What if the new headline is shorter and the layout shifts?
A: Put the button back where it was, or write the shift into the test sentence so you do not pretend the words were the only change.
Q: How do I know the new sentence won?
A: Similar vote counts, a crush rate you can cite with those counts, and comments that mention the emphasis you changed. Otherwise record no preference or a leak.