How to keep one call to action when the page wants two
A hero that asks for a demo and a free trial is two offers. Choose the action the first screen is allowed to request, and put the second ask further down the page.
Picture Chris, a founder in Seattle, arguing with a cofounder over a hero that has three buttons and no silence around them. One button books a demo. One starts a trial. One joins a newsletter that marketing promised to a partner. Each button has a stakeholder. None of them has a job on the first screen except to avoid losing the argument. The page looks busy because it is a treaty. A treaty is a hard thing to vote on. A voter can crush it or pass it. They cannot tell you which treaty clause they meant, unless a comment happens to name a button.
LandingCrush will record that preference on the homepage you submit. It will not split your live traffic between the buttons, and it will not tell you which button created a paid account. Those clicks live in your analytics. What you can do here is decide which action the capture is allowed to ask for, submit that page, and read whether people crush a first screen that makes one request. If you want to compare two requests, make them two homepages that differ only there. The sentence for that test is in how to write the one difference.
List the actions hiding in the hero
Write every action a person could take without scrolling. Include text links, not only colored buttons. "Log in" is an action. "See pricing" is an action. "Watch the film" is an action. Put the stakeholder next to each one, because the stakeholder is why it is there. The list is often longer than the team admits. A quiet text link to a webinar still asks for a decision.
Mark which actions are the same job in different clothes. "Book a demo" and "Talk to sales" might be one job. "Start free" and "Book a demo" are not. One creates an account. One creates a meeting. If both sit in the hero, you are asking a stranger to pick your go-to-market before they have agreed they care. Some visitors can do that. Many will pass the page because the page passed the choice back to them.
Mark which actions are for people who are already customers. Login belongs in a corner for them. It does not deserve equal weight with the thing a new visitor is supposed to do. Equal weight is how a hero becomes a directory.
Pick the one action the first screen may request
Choose with a rule, not with rank in the company. The rule: the first screen may request the action you want a new visitor to take if they understand the offer. Write that action as a button label you would not be embarrassed to repeat. "Book a 20-minute demo." "Start the checklist." "See the sample close." Labels that say "Get started" or "Learn more" survive the treaty and fail a stranger. Get started where. Learn what.
Say the second-best action out loud and then move it. It can live under the first proof, or in the pricing block, or in the corner as a text link. It should not wear the same size and color as the primary button. You are not deleting the business. You are sequencing it. A newsletter can exist. It should not compete with the demo in the same glance unless the newsletter is the product.
You might be thinking two buttons are a best practice because large companies use them. Some do. Their brand, their sales team, and their returning audience are not your capture. If you want to know whether one request beats two on a page like yours, submit both versions with everything else frozen. Do not cite a company you do not have data for. The Hall of Fame can show you public pages. Critique one of them with the Hall of Fame worksheet instead of copying its button count.
Check the offer line. If the primary action is a demo, the headline should make a demo feel like the obvious next step. A headline about self-serve magic sitting on a "Book a demo" button is two stories. Fix the mismatch before you submit. Voters will react to the mismatch even if they do not use the word.
Make the secondary ask easy to find and hard to confuse
Put the second action far enough down that the first screen reads as one request, and close enough that a motivated visitor can reach it. "Far enough" means it is not in the hero cluster. "Close enough" means you are not hiding pricing from someone who will not book a meeting blind. Pricing can be a link. It does not have to be a second pill button.
Write the label so it cannot be mistaken for the primary. If the primary is "Book a demo," the secondary is not "Book a call." That is the same door in a different coat, or it is a trap. If the secondary is "Read the docs," say that. Docs are a real path for some products. They are a poor primary action if your business depends on a conversation, and a poor thing to hide if your buyers will not talk to sales first.
On the phone, stack this with suspicion. Two equal buttons plus a sticky bar plus a chat bubble is four asks. Run the five-second timer from how to score a hero on the mobile capture. If you remember two actions, the sequence failed. Resize until you remember one.
Navigation counts. A bar with Product, Solutions, Pricing, Customers, Login, and a button is a second homepage above the hero. You do not have to remove the bar to have a point of view. You do have to stop adding hero buttons to compensate for a bar that already offered every path.
Submit the single-request page and read comments for hesitation
Submit the page that asks for one thing. After approval, read comments for hesitation rather than for taste. "I do not know what happens when I click" is a label problem. "I wanted the price before a meeting" is a sequencing problem. You can answer the second by making pricing findable without putting a second primary button back in the hero. "Too much going on" may mean you lost the argument and the treaty buttons are still there. Look at the capture before you defend it.
If you are comparing one button against two, keep the headline, the proof, and the picture still. A crush for the quiet page might be a crush for the picture if you changed that too. Read the count before you announce that minimalism won. The rate-reading steps are in how to read a crush rate.
The post-approval AI critique includes a call-to-action score from 0 to 10. Use it as a prompt to look, not as the vote. A high score with comments that name two competing buttons means the model and the humans are not talking about the same problem. Follow the repeated comment. The AI Panel's Click Gravity score, if you bought a panel test, is yet another column. It is a simulation. It is not the swipe. Keep it out of the button meeting or label it as a preview.
Tell stakeholders what moved, in one sentence each
The treaty breaks in private if you only delete buttons in the file. Write to each stakeholder: where their action went, and why the hero kept a different one. "The newsletter is linked under the first story. The hero asks for a demo because that is the action we are willing to learn about in this vote." People can disagree with that sentence. They cannot claim they were erased if you show them the link.
Then stop adding a third compromise button the night before you submit. The capture you upload is the capture voters see. A last-minute pill "for balance" restores the treaty and wastes the submission.
Your visitors on the live site still choose with their cursors. After you ship the single-request hero, watch the action you kept and the action you moved. LandingCrush will have told you whether voters crushed a page that asked for one thing. Your analytics tell you whether anyone did the thing. Both readings are allowed to humble you. Only one of them is revenue.
If your hero could keep only one request, which label would you let a stranger remember?
Common questions
Q: Does LandingCrush tell me which button converts?
A: No. Votes are crushes and passes on the submitted homepage. Button clicks and accounts are measured in your analytics.
Q: Can I test one button against two?
A: Yes, as two submissions that differ only in that choice. Keep the offer and the rest of the hero still, and compare crush rates only when the vote counts are similar.
Q: Where should login go?
A: Keep it available for customers, and keep it from wearing the same weight as the action you want a new visitor to take.