A game turn should be simple, easy, and interesting. It should not be a list of steps and procedures.
After complexity, this is probably the #2 criticism I have of other designers' games I playtest.
This is a huge mistake I see in game expansions. Have you ever felt like a game was simpler and better without its expansion? That extra step each turn/round has a cost.
A turn should not be comprised of "phases" unless absolutely necessary. Make a turn about a single, consequential action. Remove or minimise everything else.
Turn structure is a very high design priority. Redesign other parts of your game, in order to simplify it.
Work backwards from a simple turn structure.
Simpler turns are easier, but they're also quicker. In my experience, boring games are usually boring because of downtime, i.e. waiting for other players' lengthy and complex turns.
Doing two things on your turn is acceptable. If you can move then attack, that's fine. If you get income, then spend it, that's fine too. For more complex games, three actions is more appropriate — two choices, and one maintenance step. For simple games, the ideal number of actions is one, but I find it very hard to get below two. For simple games, you absolutely cannot exceed two actions per turn.
There's a simple test here, for you and your playtesters: are there parts of the turn that you keep forgetting, don't actually care about, or are a drag? If so, your turn structure is wrong.
In Pulsar 2849, on each turn, each player will place two dice. They'll use this to fly spaceships around, build things, and upgrade technology. This is interesting. However, at the start of each round, there's an extensive and convoluted phase where players choose their dice for the round, from a shared pool of rolled dice. Furthermore, players typically plan out their whole round during this phase, before picking dice.
If a player has to make a choice, they'll take time to think about
that choice. They'll do this, even if the choice isn't important.
Steps in a turn are bad, but steps with choices in them are worse. Try to confine the choices to one part of the turn, and make those choices important. Make the rest of the turn automatic and choiceless.
Don't go overboard in this. A simple and automatic drawing of one card per turn is likely worthwhile.
In my gangster game, drawing one card per turn gave too many cards, so I let players buy one, at the end of each turn. This step now contained a decision, which slowed the game right down. I eventually decided that, if there had to be an extra step, it would simply be to draw one card. This was too many cards, but I was determined to make it work. I made a portion of the deck super-simple, non-functional cards. These "Package" cards can be collected, and delivered to certain locations in bulk, to exchange for resources. These cards have almost no complexity, so a player having numerous packages, and one or two regular cards, doesn't overwhelm them, like a hand of several regular cards would.
All turn options should generally be put into one menu. They should not be a series of different menus of choices. "You may do A or B or C" is much faster than "you may do A or B, then you may do C or not", and offers the player more freedom. For more complex parts of a game, players might pick more than one choice, but it would still be from one menu. Giving each thing a cost works in much the same way.
In Radlands, you can put people onto the table, and you can have them use their abilities. You can use your camps, or use special actions like drawing a card, or playing a special event. However, all these things are in a single step. There is no separate "combat phase". All these things cost an amount of water to activate, and you have three water to spend each turn. Once you've spent your water, you end your turn.
Be careful of giving players options they can do on a turn, but usually don't. These always require thinking, and they can also make it unclear when a player's turn has actually ended.
Another way to simplify your turn structure is to cut your turns up into small bits, and have each bit be a turn. It's better to do half as much on a turn, if it means you wait half as long for your turn.
For simpler games, "rest turns" are fine, wherein you might just draw a new hand of cards, replenish your energy, or the like. This allows you to remove "draw a card each turn" or "gain one energy each turn" from the turn structure, which is often essential for family-weight games.
In Ticket to Ride, a player's turn is either spending cards to build "tracks" on the board, OR taking two new cards. I agree with this structure, especially since the game is fairly light and simple.
Cut out tiny tasks and choices from players' turns.
Ticket to Ride lets you take one card, put out a new one, and then decide which second card to take. You have to consider if you want the new card. This adds a second small decision and a task to the turn. I would've just had the replacement cards come out after both cards were taken.
I just want to play the game!
A player's turn should just be about doing the thing (or things) they want to do, plus minimal (ideally zero) maintenance and bookkeeping.
If there are tasks that must be done, to maintain the game state, try extremely hard to move them from the turn to the round.
Putting out ten tokens at the start of a round is fine. Having ten
turns, in which the player must also put out one token, is not so
good.
Even better, get away from having "rounds" entirely. Rounds tend to exist because a game needs to undergo housekeeping. Does your game really need that housekeeping?
In one of my prototypes, players move around a town.
The spaces in town do various things. However, to stop a player using a
spot repeatedly, and encourage them to travel more, a player puts a
"depletion" counter on the space, when they use it, so they can't use it again. After this, the
player draws a "Change" card. These cards move the non-player characters
around the board, advance the time, remove the depletion counters, and
cause larger game events to happen.
The depletion counters and Change cards here are beneficial. However,
they bog the turn down with housekeeping. A poor game designer
argues the case as to why these systems are necessary. A good designer
abolishes them, and then finds other ways to make the game work. I
removed the depletion counters, and replaced them with much rarer
"stash" counters, that add goodies to certain spaces on the board, to
encourage players to move across the board. I put the Change cards into
the other decks, so
they would be drawn occasionally, as part of players' regular turns.
Try to keep game-controlled effects out of the game. Yes, you can have "event cards" that happen periodically, or "neutral players" that perform actions, but does the game need them? If they're needed, make them very simple and quick. Far better to let a player control the "bad guy" momentarily, and use it to attack their opponents, than to have a whole process for deciding what the bad guy does.
In one of my prototypes, I decided that you could get a bad "reputation". But if you had a bad reputation, what did that mean? I could've had neutral characters who would periodically chase you, or detect you and react to you. I decided not to do that. Instead, it's the other players who can target you, and steal your points, if you have a bad reputation. It's far simpler, and requires no maintenance.
Also, beware of actions that are more complicated. It's not just the average turn length — a single prolonged turn is also bad.
In my gangster game, you go to a location, and roll one or more dice, and do their effects. Then, you can play cards. It's possible to play multiple cards in one turn. This could become too lengthy, so I declared that cards just do effects. They never roll dice.