Let the team walk its path
Every team walks at its own pace and I haven't found a way to speed it up myself, I don't think there is one regardless of whether you are an external coach or a regular member of the team. Trying to push the team will screw up the work environment, the pressure may introduce interpersonal conflicts and severe damages in the quality of the product. I mean, people will do stuff they do know it's wrong because of the sense of emergency. On the other hand trying to pull the team away will give its members the impression that they are not trusted professionals, again causing problems among people and discouraging them from being proactive and willing to learn.
The only way we can help the team progress is to help them walk their own path, removing impediments along the way, answering questions when asked to do so.
I've learned that it's not possible to develop a software product at the speed that I'd like whilst mentoring people. I have to choose either to be a trainer or to be a developer but not both things at the same time because the goals are different. When working as a trainer the goal is that people learn effectively and the best way, for us humans, to learn is by making mistakes that let us learn from ourselves. As a mentor I must combine small theoretical explanations, recommendations and practical examples with time for people to practice and to make their own decisions. They must have the chance to fail and discover. If they don't recognize the failure my mission is to provide feedback and explain why is this or that a mistake and what are the consequences. Deep lessons are learned with pain - success is fantastic but it doesn't teach. However this style of mentoring takes a lot of time, it's probably the top most time consuming way. It's a company wide commitment, an effort for everyone.
Often companies ask us to join them for a new project that must deliver value within a tight schedule and at the same time they want to build up a team of XP developers with people that have never been exposed to XP. I must say this is completely impossible in my experience. If the quality of the product needs to be excellent and deadlines are tight, the best thing the company can do is to hire a seasoned high-performing XP team whose members have been working together for years. On the other hand, if the actual goal is to build up a team, this is an investment that lasts several years. According to my experience, a group of 6 to 8 developers become an XP team after 2 years of training, guidance, reviews and practice. It requires several training courses during the various stages of the transition, practical examples, guidance and moreover it requires small failures (hopefully reversible ones). People have to have the margin to fail and learn from their mistakes. During this period some people will very likely leave the company and some new people will join.
I know the word quality is too abstract, we don't have a definition for what quality is but I am actually talking about well-known maintenance issues. A common mistake beginners make is to write brittle and hard to maintain tests that with time impede developers changing or adding features. Those tests are eventually deleted. It's waste unless you consider this to be another stage in the team's learning curve.
Nobody writes perfect code anyway, every team makes mistakes so the point is not to aim for perfection. The point is to avoid or minimize the amount of things we do know are waste, basically because we have failed in the past.
Hiring seasoned external consultants means they won't make the same mistakes they did in the past although they'll make new mistakes. Every team is different as well as every project.
As a consultant I won't prevent the team from making mistakes when the goal is to train them, instead I'll try to keep the cycles small enough so that mistakes can be fixed soon and cheap. But it's up to the team to ask for feedback and review, my job is not to control the source code - it's to support the team!
In some situations it could be beneficial that a small group of experts write the "walking skeleton" with the first group of product features in order to define architecture, style and conventions that others can follow. It serves to explore risks and exposes an example that others can follow. There are several perils of this approach though. I'd be careful with it - sorry don't feel like writing about these drawbacks now.