As a matter of fact, the 23% are pretty good compared to SupCom. Only ~10% of the players of SupCom ever created an online account for GPGnet as far as I remember. So what are the consequences?
Companies concentrating on the multiplayer part mostly (as a measure of 'copy protection') will invariably develop their game ignoring the largest part of their customers.
Releasing a game finished, without major bugs and all the planned content is crucial! While adding content later (as DLC, patch...) will bring back players and keep the online community healthy it will not reach the largest part of the customers. You can probably get away with adding features primarily aimed at the online community later (replays, modding support, new achievements, items, balance changes...) but you should have a good campaign/single player part, story, ending and features that increase replayablility from the get go.
Publishers and developers should really rethink their approach to day 0 patches! Fixing that gamebreaking bug that is a show stopper for the campaign on the release day is not good enough. 77% of the players will just uninstall your game, return it or avoid your products in the future.
AI development is important! The first impression every player will get from your game is playing against the AI. If it freezes, suicides or behaves strangely it will hurt everybody and massively decrease the replayablility of your game.
Randomly generated maps, additional gameplay modes, different storylines additional characters and difficulty levels will be more than worth their money.