You found your core gameplay mechanics, you have nice graphics, the game is running smoothly, after completing the single player campaign (or even in the middle of it), players just drop your game. Are there any ways to increase the replayability? Why, yes, yes there are.
Every single video game is based of a core loop, an action required to complete the game goal that the player will do over and over again. For platform games, it will be running and jumping, for strategy games it will be giving orders to units, for first-person shooters it will be about clicking on targets. However, the boredom specter is never far. Game designers have to come up with creative ways to entice the player to perform the core loop as much as possible. This is what I mean by replayability.
The following ways are ordered by their decreasing expensiveness to add to a game.
This is the most obvious one as it is also the most expensive feature to add to a game. Ranging from a simple highscore table to a real-time online virtual world, multiplayer features add either competition, cooperation or both between real players. Wether it is to increase its own clout or just to spend a good time with friends, players will want to play the game not just for the sake of it. No wonder it is one of the most frequent requested features on the Early Access games forums.
If you can't or won't add multiplayer features, fake it! It is less expensive than actual multiplayer features, but not by far. To be actually efficient, any AI should be sufficiently challenging and not cheat too much nor be too predictable in order to reach that goal. Sure, Artificial Intelligence is piggy-backing upon existing gameplay features, but it takes a significant amount of work to get it right. A good example is Valve's Dota 2 [fr] bots that are way more challenging than in Riot's League of Legends [fr] or Blizzard's Heroes of the Storm.
You can't grant every single change or feature request from your player, why not giving them the actual possibility to implement them? This feature is tricky because it has to be thought out from the very beginning of the project. There is very little chance to be able to add modding feature at a reasonable cost if it wasn't planned initially. However, if it was, it is relatively straightforward. Usually the main game will itself be a sort of standalone mod of the game engine. It also is a great way to foster a creative community which will indirectly increase the replayability of the game. Successful examples range from RPGs like The Elder Scrolls to strategy games like Civilization.
Random drop and cosmetic items
This may seem like a gimmick, and on some ways it is, but it is undeniably effective to sustain a video game and even bring in some revenue without having to resort to subscription or even having to sell the original game. Valve is now well-known to have pulled this card successfully at least twice, first with Team Fortress 2, informally known as Hat Fortress 2, then with Dota 2, whose set amount of heroes each have multiple skins, many of them being available only as a random drop by playing online or given as rewards for special events. Both games are free-to-play without any paywall, and yet Valve is making money, mainly out of cosmetic items and tournament ticket sales. Cost-wise, cosmetic items are pretty easy to implement in the vast majority of cases, but the sales/trade part can become tricky as with everything related to real money being spent online.
Maybe I should have added "Meaningful Achievements". Getting an achievement for completing the tutorial doesn't help your game's replayability that much. On the contrary, if you know about a rare situation happening in-game or a difficult optional goal like finishing a level in a certain amount of time, or winning unharmed, those are very good candidates for achievements. In general, I feel like their full description should be published, as secret achievements eventually get vented on wikis or gamer's forums. Consisting only of a few checks for completion added here and there, achievements are inexpensive and can usually be even added post-launch without significant game rework.
This one may be my favorite, as it is one of the least expensive ways of making your game last longer and probably one of the most overlooked. Whatever the game, most players will try to find the most efficient way of reaching the game goal. It could be figuring out what weapon suits any playstyle the best, finding the game-breaking combo, or just a preferred way of playing, even though the game offers many others ways. Adding a level where a certain game feature is disabled forces players to try something new, to get out the sweet spot they found and renew the challenge.
As a personal preference, I prefer this way instead of artificially ramping up the difficulty up to absurdity. If a combo I'm proud of isn't working on the highest difficulty level, chances are I'm going to drop the game instead of trying a new way just because the same obstacle is now that much higher. But if you present me with the same obstacle but one of the element of my combo isn't available any more, I'll much rather find a new way of overcoming the obstacle because the initial situation has changed.
As a general counter-example, Tropico 4 very rarely restricts the building selection available to the player, and instead makes up those increasingly hard (or long) to reach goals. Fortunately, the same method can usually win every single game, which made me completely ignore the Modern Times DLC that just adds new missions and a few buildings replacing the existing ones. But the same build order still applies, failing to renew my interest for the game.
While they're always not directly related to the actual gameplay, there are a number of feature considerations to be made when undertaking a video game development project. One of them is how much players will want to play your game over and over again. The different ways to achieve a good replayability depicted in this article aren't the only ways, but I feel it's a good start for game creators.