The first version of landflow could make anything. You described a site and it drew a site: a homepage, an about page, a pricing page, a contact form somewhere near the bottom. It demoed well. Almost nobody finished one.
When we read the logs, the pattern was embarrassing in its clarity. People opened the product with a campaign already in mind. They had an ad going live on Thursday. They needed one page with one offer and somewhere to put the email address. The five other pages we generated were noise they had to delete before they could publish.
What a narrow product gets to decide
A builder that can make anything has to ask you everything, because it cannot know which of a hundred shapes you meant. A builder that makes one thing can decide, and the deciding turns out to be the part people actually wanted.
Because we only make a landing page with a lead form, we can assume things a general builder cannot:
- There is exactly one conversion goal, so every section can be judged against it.
- There is exactly one form, so a second call to action competing with it is a bug, not a preference.
- The form has a destination, so attribution and delivery are part of generation rather than a later integration.
- The page is short enough to hold in one scroll, so in-page navigation is a real option and a sitemap is not.
Each assumption removes a question from the brief. What is left is the part only you know: what you sell, who to, and what counts as a win.
What we gave up
We are not a website builder and will not become one. If you need a documentation site, a store, or twelve pages of company history, we are the wrong tool and we would rather say so on the first screen than three hours in.
A builder that can make anything has to ask you everything. A builder that makes one thing can decide, and the deciding is what you actually wanted.
The test we use now
Every feature proposal has to survive one question: does this help someone put a page behind an ad this week? Section reordering survived. Custom code in the head survived, because site verification needs it. A blog engine did not. A component library did not. A second form on the same page very much did not.
The product is smaller than it was a year ago and people finish pages with it. That trade has been worth making every single time.