Years and years ago, when I still suffered from the delusion that I was a eurogame designer, one of my games was picked up by a small firm who never quite got around to publishing it. While that experience wasn't entirely a pleasant one, they did have several playtest groups at their disposal, and the feedback from those playtesters was generally useful and constructive, and made for a game that would have appealed to a broader audience than my original design if it had ever made it to market.
Every time there was feedback, the publisher amended the ruleset, often adding a new example to help clarify this thing or that one. The weird thing though is that as development continued, the game got harder and harder for each new group of playtesters to understand. They were missing fairly simple and obvious rules, overlooking information that was plainly stated. The publisher's solution was to restate the rule a second time for emphasis, or to add yet another example. So you would get stuff like:
The player with the First Player marker goes first in each round, and play passes clockwise. When it is going to be the First Player's turn again, the marker is passed counter-clockwise and a new round begins. Therefore the player who was last in the previous round will be first in the next.
Example 1: Pat, Alex, David, and Susan are seated in that order and Pat is the First Player. Alex has the next turn, followed by David, than Susan. Then, Pat passes the First Player marker to Susan, and a new round begins: Susan has the first turn, followed by Pat, Alex, and David.
Example 2: If David, Alex, and Susan are seated in that order, and Alex is the First Player, play would pass to Susan than to David. Then, Alex would pass the First Player marker to David.
In this way, the ruleset ballooned in size, until the book for this simple, streamlined euro-style game was longer than most wargame rulebooks. Then the publisher brought in a copy editor, who had more questions, and the publisher's answer was to provide more examples, making the thing even longer and more convoluted. At that point, the rulebook was over five times as long as it had been when I had submitted it.
I had been trying to fight against this bloat during the two years that the game's rules were under revision, but my fighting was half-hearted and wimpy: I was a neophyte game designer, not at all confident in my abilities, working with an established publisher who seemed to know what they were doing and what they were talking about. When they told me that the problem with most rulebooks is that there weren't enough examples, that made perfect sense to me. When I tried to argue against the inclusion of a new example, I was rebuked: "Maybe you're right, and ninety-nine percent of gamers don't need the example. But if we put it in there, it will help that other one percent, and besides, it doesn't hurt putting it in there, does it?"
But it did hurt, because it made the rulebook bloated and indecipherable. A new playtest group was handed this version of the rules, and they just couldn't make heads or tails of it. They got almost every rule wrong, and it was just an awful two-hour slog. The publisher was present at the playtest but didn't correct them, for doing so would apparently undermine the sanctity of blind playtesting. He was extremely discouraged, and wrote me a likewise discouraging email recounting the session. "They couldn't understand [concept] at all, despite the fact that we have six examples."
But really of course, they couldn't understand the concept because we had six examples. It made the concept and the game sound more complicated than it really was, and at this point, tired of meekly protesting, I just went ahead and rewrote the rules from scratch, expressing each concept in its purest, simplest form, with only a handful of examples. The publisher, the copy editor, and the playtest groups all agreed that it was vastly easier to understand. It wasn't quite there yet - it did in fact need a couple of additional examples - but it was at last moving in the right direction. The game might not have gotten published, but that wasn't because of any difficulty comprehending the final ruleset.
Examples are like herbs and spices: they wake up the palette, highlight the flavors, and bring the whole meal together. But if you just throw a whole mess of them in your pot, it's going to overpower the main ingredients. And just as you have to be careful with what spices you're going to use - I'm looking at you, fennel! - you have to be careful which rules you provide an example for. You don't need to provide an example for things like moving from one hex to another, or advancing the Game Turn Marker. You probably should provide an example for combat, or to explain a novel concept that's harder to grasp or is easily overlooked.
Example of Play
But what if everything in the game is new and unusual? Personally, I think the best solution in that case is to just write one long extended example of play that you can shove in the back of the book. It's actually hard to do this well, because you want your sample turn to touch on as many of the mechanisms and concepts as possible, cramming them much closer together than they would be in " normal" play, without the normal room to breathe, and yet the whole thing still needs to feel organic. It's much easier to come up with examples piecemeal for each mechanism than to string a run of them together into a coherent whole, but it's also the worst thing you can do in that situation, because the examples will be constantly butting in and interrupting the flow of the rules, bogging the whole thing down.