Scaling is the process by which you make your game work properly (and hopefully well), regardless of how many people are playing.
A map which is suited to five players, might be too big when the game is played with two. With more players, there might not be enough resources in play for everyone to get enough. In a cooperative game with more players, the game objective might be too easy.
It doesn't matter if the game plays differently at different player counts, as long as it's still good.
Doing nothing (or just a little bit) can be a very acceptable solution to scaling issues. The best scaling is no scaling at all.
If your game has a board, a basic solution is two have a two-sided board — one for 2-3 players, and one for 4+ players.
Another option is simply to have an extra "sliver" or two of board that attaches to the side, with some extra spaces on it. This is my preferred option.
The boards in Caverna are cut up into slivers. You flip or remove some of them. There's a combination of board slivers for every player count from 2-7.
If your game has cards, having to fish through the game's decks, to remove all the cards for other player counts is onerous. If you have to alter the decks, just make it a very small number of impactful cards being added or removed.
As an alternative, you can simply add special rules to cards, such that they do something extra or different, if you're in a game with a specific number of players. These are best in games that have "event cards", where a bit of extra text is okay, and the card affects everyone. These variable cards require extra words, but they don't require extra components, or that the players sift through the cards before each game.
Overt scaling abilities that say something like "gain wood tokens equal to the number of players" can work, but are a bit "dirty laundry".
In my gangster game, players attack all their opponents simultaneously, by rolling custom dice. In a 5-player game, you have four other people attacking you, and this gets lethal too quickly. I made two faces of the red die a "grenade" ability. This ability causes opponents to lose 1 health if there are 4-5 players, 2 health if there are 3 players, and 3 health if there are 2 players. This was a very easy way to scale the game. The games with more players are still more lethal, but it's still reasonable.
One strategy is an "option hierarchy", whereby actions range from good down to bad, regardless of the number of players. With less players, the players will still compete for the best ones, but with more players, the lower-value options will come into play.
Aim your game at its most common numbers of players (likely 3-4 for a 2-5 player game.) Some things might be weak in a 2-player game, but very strong in a 5-player game. As long as nothing becomes either unusable or overpowered, it's fine.
Ideally, you just want your game to be playable without any overt scaling. Some games can simply be played perfectly well with different player counts, with no special rules.
The game length shouldn't be multiplied by having more players. Two hours is fine for a two-player game, but players will simply not play the game with five players if it's going to take five hours. You need to drastically cut back the game end condition, or number of rounds, even if it impacts the game.
If a game ends when its components run out, that can be a great way to scale the game length. Just use the same number of components, regardless of player count.