Turning Bads into Goods
When starting a new raiding guild, you're eventually going to hit that chasm between the 10 and 25 person roster sizes. Staying above 13 and below 22 active raiding members for too long can be dangerous - you can't offer everyone a spot every raid, you can't run two 10 person teams, and you can't do 25 person raids without bringing in a few PUGs. If you don't push through quickly enough, you risk losing members and being forever stuck as a 10 person guild.
As a guild leader, the need to build up your roster may be stronger than your principle to only bring in qualified raiders. Is it better to have 25 warm bodies, some of whom aren't the best at their class, or to have 17 excellent raiders, 7 of which have to sit out each raid? Pragmatism will probably win out here, and you'll end up with a mix of skill levels in your ranks.
The question now is how much time you're willing to spend to turn the bad players into acceptable (or even exceptional players). You have a few options:
Weed Them Out During the Trial
The most obvious answer is just to never bring bad players in as full members - be very open in terms of who you will bring in as a trial, but take a firm stance on the performance that any given role has to meet in raids. If you can't meet that level of performance, you'll get booted at the end of your trial and someone new will replace you. It's entirely your responsibility to improve your performance.
This is the technique that most guilds tend to use, and when you're already at 25 person strength it's a reasonable approach. When building up to 25 person strength, you can waste a lot of time here though - you have to give people a few raids to find their groove, so you need a trial period of at least two or three weeks. Assuming your trial members are allowed to receive loot, this means that you may invest a few weeks of time and give several pieces of gear to someone who you end up booting out the door. Do you stay principled and boot them if they can't put out the performance, or is there a point where you start thinking "the guild has now given this person 8 upgrades, and I don't want to lose that gear"? Decide these things ahead of time.
Read the Fine Manual
Thanks to the proliferation of WoW sites and blogs, there is a wealth of information to be had on how to be a better player for any class and spec you can choose. If someone isn't performing up to spec, you can just send them to a good source of information and hope that it sinks in.
Remember though that some people just aren't book learners - they have to be shown how it's done rather than poring over pages and pages of forum posts.
Assign a Mentor
If you have a willing and capable member of the same class, you can use a buddy system. Pair up a the poor player with the good player. Have an initial chat between all three of you (unless the job of improving performance belongs to a class lead in the first place) and discuss the issue. Have the good player analyze every aspect of the bad player's game - rotation, gear, enchants, gems, spec, movement, non-combat roles (crowd control or dispels), and especially the reasons that they die if it's to anything but a wipe.
There are plenty of tools out there to perform this analysis, so make sure that the mentor is familiar with them - if your guild always uploads a combat parse then you can use that, but if it's a sporadic thing for the guild then make sure that the mentor is doing their own combat logging.
A side benefit of this approach is that if assigning a mentor forces the mentor to learn new tools and new levels of analysis, they may be a good candidate to be a raid leader, as the skills required to analyze the performance of one person aren't very different from the skills required to analyze the performance of the raid.
Make sure that the mentor discusses performance after every raid (or the next day). Don't let several raids go by without the bad player getting some solid feedback. Even if it's just "your performance was better than last week, but I need you to tighten up your rotation to reduce your DPS idle time", that's something that the student can focus on. Every time you do analysis, try to pick the worst aspect and get them to improve that - don't give them a list of ten things they need to work on next raid.
If you don't have a capable member of the bad player's class, you can do this yourself - but it helps to have more than a passing familiarity with how that class plays at 80. Unless you're a serious altoholic, the level of specific analysis you can provide will be limited.
"Worst In, First Out". Recruiting needs to be a continual process. Assuming you achieve a moderate level of progression (possibly with your good members carrying your bad members), recruiting is going to get easier. You will see more qualified candidates. When you get a stellar trial member in, you do have the option of giving their raid spot to someone who isn't performing. You might even kick the worst members from the guild entirely. It's a bit cut-throat, but perhaps that's how you want to run your guild.
If you choose to do this, be upfront with your members. Don't stealthily just stop inviting bad players to raids. Make it part of your guild policy that your raid spots go to the best performing members regardless of how long they've been with the guild, and that everyone has to perform or get out.
Don't Forget Gear
When evaluating any player, don't forget to set your expectations to match their gear level. You don't want to dismiss a great player because their performance isn't where you want it to be only to find out that they were pumping out as much as was reasonable in their gear. If you choose to pick another established member whose gear score is similar, make sure that the skill level and spec is also in the same neighbourhood before you compare their performance. Class leads can be helpful here if they know their stuff, as can the various class spreadsheets that float around the net. Remember as well to adjust your expectations as your members gear up (or don't if there is a streak of unlucky drops for a class).
Here's hoping that you spend as little time as possible in the 10-25 chasm.
Until Next Time