How to turn LandingCrush vote data into a homepage decision
How to read crush rate, vote volume, velocity, comments, version history, and the AI critique on LandingCrush, then choose a homepage direction.
LandingCrush vote data can support a presentation decision: which headline to keep, whether the hero screenshot earned more crushes than the illustration, whether the call to action is passed more often than it is crushed. Write that decision before you open the dashboard. “We will ship variant B’s headline if its crush rate is higher after both pages have a similar number of votes, and if comments do not reveal a different problem” is a rule the product can inform. A pricing change, a hiring plan, or a revenue target is outside what a swipe records. The about page says the tools do not guarantee a conversion outcome.
Start with the creator dashboard
Sign in and open /dashboard. Each submission shows status, total votes, and crush rate. Status is pending, approved, or rejected. Pending pages are still in moderation, so they are not a live feed result. The analytics view adds weekly votes and velocity. The data model describes velocity as votes per hour over the last 48 hours. When votes exist, a daily chart shows whether they arrived in a burst or across days.
Crush rate is crushes divided by total votes. Read it beside the count. A perfect rate on three votes is not a decision. A mid-range rate on a large count is a preference share, and it is still not a conversion rate. The glossary calls a crush rate above 70 percent strong. Use that as LandingCrush’s label for a high preference share, then check whether the other version in your comparison is lower by a gap that survives similar volume.
Weekly votes and velocity tell you if the page is still being seen. A strong rate that stopped accumulating is a stale snapshot. A page approved yesterday should not be declared the loser against a page that has been in the feed for a month, unless you limit the comparison to a shared window or run a duel. The public leaderboard — All-Time, This Week, Rising Stars, and industry filters — shows other approved homepages. Use an industry filter as context for the category. It is not a benchmark that guarantees your rate.
Let comments explain the rate
The detail page lists comments. The binary vote says which way the swipe went. The comment says why, when someone typed. Look for repeated subjects: pricing, the headline, the screenshot, logos, or “I don’t know what this does.” Three comments with the same confusion are more actionable than a two-point gap in crush rate. If you only read the percentage, you can “win” a test and ship a page whose comments all say the button is unclear.
Roast mode asks the voter for a critique. Rate mode is the professional assessment style. Both still store a crush or a pass. Put the repeated comment in the decision note next to the rate. Blind Mode also changes the interpretation. Votes can be cast with the company name hidden. If the question is whether the layout stands without the brand, those votes are the stricter read. If the question is whether a known name is doing the work, say that explicitly instead of folding every vote into one story.
The dashboard can break voters down by persona role and company size when those voters filled the fields in. Roles include founder, designer, marketer, developer, and investor. Treat the breakdown as a description of participants who opted in. An empty breakdown means people voted without a persona, not that a segment rejected the page.
Compare versions, not memory
Homepage records support a parent id and a version number, so a later iteration can be tied to an earlier submission. The free plan includes one active homepage. Creator Pro, at $19 per month or $190 per year, includes unlimited submissions, which is how you keep the old and the new page both measurable. Compare approved pages with approved pages. A pending upload has no feed result yet.
For a bounded clock, start a Creator Duel from an approved homepage. The duel runs 48 hours and returns a link. That is the right artifact when two people want the same window, rather than two pages that collected votes on different weeks. Line up vote count, crush rate, the one change you intended, and the comments about that change. If the counts differ sharply, the result is incomplete. If the rates are close, there is no preference to obey: keep the version that is simpler to build. If one rate is higher, the comments agree, and both counts have passed a small sample — the product waits for 50 real votes before it scores a panel prediction against the community — ship that presentation.
Keep the AI critique beside the vote
After approval, LandingCrush queues an AI design critique. Scores run from 0 to 10 for hero clarity, call-to-action effectiveness, visual hierarchy, and mobile responsiveness, with notes and suggestions. Pro accounts go to the priority queue. The analysis can store a category average of AI overall scores for other homepages in the same industry. That average is about the model’s scores. It is not a crush-rate average.
Read a low hero score as a hypothesis. Confirm it in comments or in the next version’s crush rate. If the model says the button is weak and the comments never mention the button, do not rebuild the button only because the score moved. The methodology page says panel personas are fictional and that human votes win when they conflict with a panel prediction. Use the same rule for the post-approval critique. The separate AI Panel Test, at $29 or $49, is optional and labeled a preview. Archive its verdict next to the swipe result and write down which one the decision follows.
Write the decision in five lines
- Question: which homepage presentation you will ship.
- Evidence: crush rate and vote count for each version, plus the duel result if you ran one.
- Reason: the comment that showed up more than once.
- Set aside on purpose: AI score gaps the comments did not support, and replies from a friend poll.
- Next check: the live-site metric you will watch after launch. LandingCrush does not collect that metric for you.
If line two is empty because the page is still pending or the count is tiny, the decision is to wait. If line three is empty and the rates match, the decision is that the feed showed no preference. Both are valid uses of the data. A homepage decision includes the choice not to change. Share the public page, or a vote-count embed, after that note exists, so the badge is a record of votes received.
Teams that skip the note tend to remember the percentage and forget the comment. The percentage chose a direction. The comment tells the designer what to change. Keep both, then verify the shipped page with the analytics on your own domain.
Common questions
Which LandingCrush number should we act on?
Act on crush rate together with total votes, and only for a presentation choice. Weekly votes and velocity tell you whether the rate is still current. Comments tell you what to change. AI scores are a separate critique.
What is a large enough sample on LandingCrush?
The product waits for 50 or more real community votes before it compares an AI prediction with the community outcome. Below that, describe the result as directional. Also avoid comparing variants with very different vote counts.
Where do I see the vote data?
The creator dashboard at /dashboard shows crush rate, total votes, weekly votes, velocity, a daily chart, persona breakdowns when voters set a role or company size, and the AI scores. The public homepage page shows the preview, comments, and the critique.
Should a higher AI score overrule a lower crush rate?
No. LandingCrush treats community votes as ground truth when they conflict with an AI panel prediction, and it labels panel output as a preview. Use the same rule for the post-approval critique.
What decision is out of scope for swipe data?
Pricing changes, hiring plans, and revenue targets are out of scope. The vote supports which submitted presentation participants crushed. Your own analytics have to answer what customers did after you shipped it.