This is an article of general advice for designing cards and other game objects.
There's huge design space in cards that can only be used when a condition is met.
The Radlands camp Nest of Spies lets you do some damage, for the cheap price of one water. However, you can only use it if you've put two people into play on the current turn.
If you're not using conditional effects, you're probably ignoring about half your game's design space. It's worth going through all the possible conditions in your game systematically, as I've found that conditional cards aren't ones you'll naturally think of, when you just design a whole lot of cards.
Some cards require a condition to be met, to be useful. If a player doesn't have control over that condition, you can give the card a second effect that helps create the condition.
The Radlands card Zeto Khan lets your Event cards happen instantaneously, rather than being delayed. This is fun, but what if you don't draw any event cards? I gave him a second ability, that lets you draw three cards then discard three. This gives you a very high chance of drawing event cards.
A good way to add a lot of strategy to cards is to create a downside, or undesirable consequence. Make cards that also affect (negatively) your own stuff.
Let the card do one of two (or more) things. This is the perfect way to merge cards with very fringe effects, or that are often useless.
Go through all the rules of the game, and make cards that break them.
Many cards have an ongoing effect, and are thus useful at the start of the game, but become useless later on. If your game has a "points" system, give cards with an ongoing effect a high point value. This means players will also want to play those cards at the start of the game, for their effect, but also at the end of the game, for their points.
Give the player a way to use unwanted cards. This also lets you give out more cards than the player needs, which increases choice.
In Radlands, cards can be "junked". There's an icon in the corner. You discard the card, and do whatever that icon says, rather than actually using the card.
In my gangster game, to play a card, you had to also discard some number of other cards from your hand.
In another of my prototypes, a few actions simply require that you discard one card. It's not much, but it makes every card useful.
Every card should relate to its context. With some situations or other cards, it's more (or less) useful. The player should have to evaluate the card, each time they draw it.
In Radlands, you choose three Camp cards. They do something, and also give you a few cards for your openinng hand. Obelisk doesn't do anything, but gives you an extraordinary three cards. Obelisk did have a downside, though. It began the game already damaged. It was a strong camp, which was fine, but it had no context. It didn't matter what other camps you had. You always took the Obelisk. It was still a fun camp though, so I changed it to destroy itself. Now, it's a weak camp, but an amusing curiosity that you need to evaluate each time you draw it.
There should be no cards you always (or never) play.
However, the player shouldn't just be able to reduce this analysis down to a very simple formula.
My gangster game had a Machine Gun card. It cost some money to play, but would increase all your future damage by 1. What's the strategic analysis created by the Machine Gun? It boils down to whether the game is going to end soon, or go on much longer. If the game is about to end, you don't play this card. If the game will go on for a while, you do play this card, as it's a worthwhile investment. This card's strategic analysis is too easily resolved to a simple question, and the card was culled from the game. (If every card was long-term like this, this card would be appropriate for the game, as you'd have to compare them all with each other. That would be an interesting choice.)
Choices that simply resolve to "am I winning or losing" should also be culled, for the same reason.
If a choice can be made now or later, both should be viable choices. There should be no card that you always play immediately. There's a lot of strategic depth in choosing timing, without added complexity.
My gangster game also had a Tank card. This card was like the Machine Gun card, but it instead doubled the amount of damage the player did in future. I made Tank work, by making it destructible. It had its own Health tokens, and protected the player's health. The Tank was discarded when it ran out of Health. This new Tank now had a second purpose — protecting the player, but the card's main ability now required much more strategy. You needed to choose when to play the Tank, because if you played it in the wrong context, other players would pummel you before your next turn, destroying your Tank before you got to use its double damage.
My gangster game also had a "you've found some money" card. You played it, and you got a dollar. There was never any reason to hold this card. You'd always play it, and always play it immediately. The card it became was Kidney Donation. It gained you two dollars, but also cost you two health. This card was super-simple, strategically interesting, was played only some of the time, was usually not played immediately, and was also humorous and relatable.