My criteria levels have been changing lately. I realize I used to make it pretty easy for the dog, any change in criteria would be slight and very understandable. I guess I thought they weren't smart enough to get it if I challenged them more. Thus, Storm (my grown up dog) evovlved into an intelligent, but not very hard working dog. Training took time, but we got there.
With Orkan (my puppy) it's a different story. He's very intelligent and very impatient, so every bit of training has happened faster. He has already taught me a lot about training, so I'm looking forward to what I hope will be a long life of learning together with him!
When the dog understands what you want, you raise your criteria to get closer to the final behavior, right? How fast? And how big are the gaps between the stages supposed to be? I've found that the answer is usually "faster and bigger than you think". The dog might get confused at first, when your'e no longer satisfied with what you just rewarded, but I think one of the things that characterizes really successful dog trainers (in shaping that is, let's not confuse with luring etc), is that they make progress faster than us mortals. And they have dogs who keep working even when the reward disapperars for a while.
Of course progression comes from a number of things, such as timing, understanding what really goes on in the dog's brain when learning, foundation trainig etc. (that could be a long list!), not just challenging criteria. But it's part of it.
Then what if the dog fails to meet the criteria? Within certain limits that's a piece of valuable information transfer. If the dog understands the situation (through good foundation training and gradually increasing difficulties) he'll keep working, because he knows he's on the right track since you just rewarded something similar. Then give the whole jackpot when he finally gets it! In fact, failing is one of the things I'm learning to love. Instead of just giving the dog info on what's right, some info on what's not right will narrow his search area for the perfect behavior down to a minimum.
We want something like 25% mistakes! If the dog has succeeded on the last seven tries, it's time to make it more challenging! For example when weave training, if the dog succeded on all the entries you have asked him to do, give him an extremely difficult one. If he succeeds, that's great! That means your dog has a very good understanding of the behavior. It also means you didn't need to spend all that time working on easy entries. That's a valuable piece of information for you... If he fails, keep working. Info on what is not the right entry will also be a valuable lesson to him. We want them to know that regardless the situation, there's only one option entrywise, don't we?
Go out, make your dog work, and have fun training!