The deciding question
How many distinct audiences do you have, each needing a different next action? One audience with one action is a landing page. Several audiences, or one audience that genuinely needs to research before committing, is a site.
The common mistake is confusing services with audiences. Five services sold to the same buyer, for the same reason, is still one page, the buyer does not want a menu, they want to know whether you solve their problem.
The opposite mistake is building a site because it feels more legitimate. A thin ten-page site is less convincing than one page that answers everything, because every page you add is another page that has to be worth reading.
Count audiences and actions, not services and ambitions. The number is usually smaller than the org chart suggests.
When one long page beats five short ones
Trdelnik had two audiences that look incompatible: an investor evaluating a bakery franchise, and a customer who has to want the pastry enough to queue for it.
Splitting them would have lost both. Instead the investment case runs at the top with hard numbers, and the appetite case runs underneath in the brand's own voice, down the same scroll.
It works because the two audiences overlap more than they appear to. The investor's real question is whether people will queue, so the appetite half is not a distraction from the investment case, it is the evidence for it.
The upgrade path
Build the landing page so it can become a site. Real sections that could each become a page, a navigation structure that can grow, and content held in a structure rather than baked into a single layout.
The cheap version (one flat design, everything hard-coded, no content model) cannot be extended. It can only be thrown away, which is how a landing page becomes an expensive rehearsal.
Decide the shape of your content even if you only have one of each today. It costs an hour now and it is the difference between growing a site and starting again.