There's a couple different ways you can approach this issue, what you pick generally will depend on your story and what reader impressions you're going for. The main thing you want to bear in mind, is do you want this game to be focused on the reader's first-time experience, or do you want it to have a lot of replay value?
In games without scripting (which is too complex for me to go into here), there are generally five different types of links you can give people:
1) Meaningful choices. Each time the player makes a choice like this, they send the narrative in a completely different direction. A game with only choices like this gives the player a lot of flexibility, but it would need to be on the shorter side in order to not be a wall of text on every page. Good for replay value, neutral for first-time reading.
2) Save-or-die choices. Each time the player makes a choice like this, they either die and get an endgame link, or advance the narrative. A game with only choices like this would be extremely linear. Neutral for replay value, good for first-time reading.
3) Flavor choices. These choices do not affect the narrative, but give flavor description or text depending on what the player selects, before routing them back to the main path. A game with only choices like this would be extremely linear. Bad for replay value, good for first-time reading.
Then, there's the two non-choice types of links:
A) Single-link choices. This is when there's only a single link on a page. Authors often do it to show the passage of time. A game with only links like this would just be a regular story.
B) Optional informational choices. Clicking this link routs the player to a page with extra background information, and then sends them back to the storygame. A game with only links like this would just be a worldbuilding infodump.
In general, you want a good mix of these choices. I can't advise you what balance to pick without knowing what your story is supposed to be like, but here's a few examples of storygames on the site that use these choices in different ways:
Eternal & several other EndMaster works: Mostly #2 choices with #1 choices at critical moments, with As and Bs spliced in here and there. The length of these games limits the amount of influence the reader has over the narrative, but adds a lot of replay value. This format seems to be the most popular on this site, especially for more literary games.
Dreamtruder: Mostly #2 and #3 choices, with a few #1 choices at critical moments. This works well with the game-like style of the game, and the three parallel narratives it offers, though the existence of a main branch limits replay value.
The book of vanishing tales: #3 choices for almost all of the narrative, though these choices do ultimately influence the major #1 choice at the end. The whimsical and narrative nature of this story works well with this structure, and the variety of choices and endings avoids common replay value problems with this setup. The Price of Freedom has a similar structure.
Unbroken & several other Ninjapikta works: A roughly even mix of #1 #2 and #3 choices. Though slightly limiting replay value, this works well to give the reader an expansive and exploratory first-time experience.
Detective 1: Blacksea Island: Almost all of the choices are #3, but this works very well for this game since it's all enabling the player to solve the mystery. At the end of the game, there are a series of important #1 choices. Replay value is extremely low, since it's a mystery game.
Hastings 1066: Aftermath: A nice mix of #1 #2 and #3 choices that expands the game and gives it a balanced feel while still offering the player significant agency, and replay value.
The path of death & several other Steve works: Mostly #3 and #2 choices, with #1 choices at critical moments. This makes for an expansive and immersive first time experience, at the cost of some replay value.
And for an utterly shameless plug:
Diplomat: Mostly #2 choices with #1 choices at critical moments, and #3 choices in a few places.
In conclusion: Pick a target audience, decide on replay value vs. first-time experience, and look into works similar to the one you're trying to create handled the issue.
