Someone recently asked me for information on starting new consulting engagements. A few years back I published Sean Doran and Scott Conley's thoughts on Being a Lead Consultant. Sean and Scott's list is great for any lead consultant, and the advice applies well for the lifetime of a project. I considered sending that list to the person looking for new engagement advice, but I'm not sure that list would be the best place to focus my attention at the beginning of a project.
The beginning of a project is a special and dangerous time. You get a mix of (at least) optimism, concern, and freedom. There are a large number of ways the project can go, and as a lead consultant you'll play a major role determining it's outcome. The specific question I recently received was: did you have a process you followed when new engagements began. The remainder of this blog post contains my (slightly edited) response.
I found being a lead at ThoughtWorks to be a nearly impossible balancing act. It's possible we didn't have a process because we weren't organized enough, but it's more likely that there's no general formula that works.
What we ran into constantly was what we called the "enablement versus delivery" issue: If you're teaching client devs (enablement), you really don't have time to meet delivery deadlines; conversely if you're delivering software you rarely have time to teach. How to balance enablement versus delivery is something that varies based on the client's skillset and the ROI of the software being developed. What makes it worse is that the client often thinks they need one thing, but if the the company didn't have issues you (likely) wouldn't be there. Sometimes the issues are with the stakeholder and they'll tell you to focus on the wrong thing, and sometimes the stakeholder knows the deal but the other people you are forced to work with are the problem.
Often it's a race to figure out what the client needs and get them to agree to fix it, before you've lost too much good will.
Another issue is, most software is worthless within a few years, and the only way to break a perpetual cycle of mediocrity is to level up the people and processes. This can lead some people to think that delivery of a specific piece of software is secondary to improving process and people. There's a major flaw to that approach though; a non-TW consultant once said that he feels guilty about his job, because when he helps people become better the most talented always end up leaving the (suboptimal environment of the) client. You can help people improve and install good processes, but when the good people leave and the remaining don't understand the foundations of the process, you end up with not enough talent and a process that's loosely followed and for none of the right reasons.
That said, you can't focus exclusively on delivery. I once led a team that beat all deadlines, created a great piece of software, and provided a great deal of content for Martin Fowler's DSL book. It was a "huge success" until the client devs took over, couldn't maintain it, and wrote their own version that had 20% of the functionality we provided. The software we wrote was classified as a "proof of concept" and thrown out.
I was somewhat oblivious to all of this in my first few years at ThoughtWorks; I would work with the talented clients while isolating the less talented. Eventually you find out that a stakeholder will usually fire you before their worst employee, regardless of how obvious it is.
If I were going into a new engagement these days, I would split my time between training and delivery initially. After figuring out which is the bigger problem, you can spend more or less time on training or delivery.
I would also keep a spreadsheet with every client employee and their talents; every one of those employees will impact your success. If you find what they're good at and get them doing that, you may have found an ally and advocate. Every employee that has no talents listed in your spreadsheet is not only slowing you down, but is also likely (consciously or unconsciously) sabotaging you in every discussion you aren't a part of. Assume you cannot get rid of them, and don't bother trying; spending political capital managing client staff is a bad investment. Keep them on your sheet and make finding their talent a top priority.
Good luck, it's not an easy gig. Then again, if things don't go well you can always move on to another client. That was what always kept me sane. I always did the best job I could, but I also knew if I failed at an impossible task it wasn't the end of the world. There's an endless stream of impossible tasks available.