I've been off-and-on for a while now, more off than on, lately… between RL work and Star Trek Online, not a ton of extra time... but I still have the COA hak downloader on my desktop for that moment when I get an itch to play again. I've been guilty, in the past, of being the guy that stirs the negative anti-DM pot... and have left for an IG experience that caused RW tension. Still, I do always come back because, well... Arabel is the closest we have to what I would build for myself if I had the time and skill to do so.
Having said that, there are a few things I think could be done to retain the players that remain, and to get new blood to fill the server back up. Some have already been mentioned, but are worth repeating...
Advertisement. This is your number one key to success and the ONLY thing that can be GUARANTEED. Everything else is nothing more than suggestions, take it or leave it. But advertising is a MUST. Just as in sales, it's a numbers game. For every 100 eyes that see your add, 10 will log in... for every 100 that log in, 25 will stay logged in after the initial introduction... out of every 10 that stay that first time, maybe 1 or 2 will return.... The more eyes you get in front of your ad, the more people you'll have playing. 1000 pairs of eyes, that's 5 new players.... 5000 pairs: 25 players. I would even go as far as to suggest paying for it... FP "sponsored" posts are only like $5 or something, relatively cheap. You can get your 5000 eyes right there. Do that once a month... who knows.
Consolidate maps. Fewer maps = fewer empty spaces = more crowded places = more player-to-player interaction. With the exception of Arabel itself, I'd say no more than a single 32x32 map per major feature. Inmersea, 1 map. Hullack forest, 1 map. King's swamp, 1 map. But at the same time, don't hide features inside a bigger map. Eveningstar should have it's own map area (albeit smaller than 32x32) and not just 3 or 4 buildings placed inside the High Road map. For the city itself, since it's the main focus of the setting, I'd suggest 3-6 maps, depending on the size of the areas. Most "dungeon" areas should be a single map... two if it's a big area. Expansive places like the Underdark should be maybe 3-5 areas... You can use creative placement of doors, area transitions, and multiple elevations to make each map seem bigger than it is... but the actual map areas themselves need to be limited.
More to do inside the city. It's the primary focus of the server, it IS the setting. It should be the primary focus of most plots and story arcs... not just a way-point to stop between adventures elsewhere. This will have the additional benefit of having more to do closer to the starting areas so that potential new players can jump right in and find something to catch their attention immediately.
This next suggestion, I can't stress enough. It took me a very long time to learn this as a DM in my own tabletop games: DUNGEON MASTERS MUST LEARN TO SAY YES TO PLAYER REQUESTS. The whole idea of submitting an application for a character background had been turning people away, and will continue to turn people away in droves. And you won't know that you've lost a player. They won't submit an app. They simply will say "EFF that" and go somewhere else, you'll never see them and will never know how many players you could have had. If a player can't play what he wants to play on your server, he WILL play it on another one. Tell one of COA's players 'NO' and you are giving some other server a brand new player. IF a player concept, or background, or special permission, or whatever causes a problem... deal with it AFTER the problem has been identified--don't deny the player his fun just because something MIGHT become a problem.
Speed it up! A week of real-time is a near-eternity in game-time. Player comes up with some plot idea, if he doesn't see results, he will drop it and leave... but not before complaining about DM favoritism and lack of attention to anyone who will listen. Player plots should be acted upon in the same session in which the DM becomes aware of the player's interest. Progress should be made within a few days, and conclusion for most plots within a week or two. Faction plots should last between 1-3 weeks, maybe a month if it's a real big one. Big server-wide world-changing plots shouldn't take more than 2-3 months.
Focus on the solo's. Yes, group interactions catch DMs attention. You want player-to-player interactions. So when someone gets a group together, sure, by all means, pay attention to that group... but they're already a group. They've already created player interaction. Adventure and intrigue already exists. That solo guy... he needs help. Give him a REASON to interact with others. Nobody can intrigue by themselves. The group will appreciate DM attention, but the solo guy NEEDS DM attention.... or he will leave.
Change it up. I'm aware of how time consuming it is to create new quests. Writing, designing, scripting, balancing monsters and treasures... it's a lot to do. But you can reuse and recycle just about everything. Take Myron's cave, for example? Its a classic. Everyone knows it and could probably solo-run the whole thing blind-folded. Maybe it's time for Myron's friend to finally get rescued for good? You can then use that quest with new NPCs on a new map, a few minor dialogue edits... and voila! You have yourself a brand new never-before-ran quest with minimal effort. There are literally THOUSANDS of published modules and side-quests that can be quickly edited and adapted for use by CoA with 90% of the work, including plot and balancing, already done. All you have to do is build the map, type in the dialogue, and fill it up with scripts and pre-balanced monsters & treasure for CoA's already existing spawn and treasure tables.
Documentation. As a new player, I want a one-stop-shop. Single-click solutions for all my character creation options, hak-paks, setting background, server-specific backgrounds... if I have to spend 3 days scanning through ten years of forum posts just to get an idea of what the server is about... I'm going to log out permanently when the server doesn't live up to my expectations and wow me right away after I took all that time and effort to figure it out.
Cater to the casual. Most people who might be interested in playing have real lives to live. Work, family... We might only have 20 or 30 minutes a day to play. They don't want, and should not have to, take a week or more to earn enough wealth and experience to raise to a "survivable" level. Sticking with the level 10 soft-cap (which I REALLY like, BTW) progression should be very rapid from 1st to 3rd... Easily raising a level in that single 20-30 minute play session. If someone has the time to spend an afternoon playing... they should be able to make 3rd of 4th in that first day. Then it should slow down until about 5th or 6th (which IMO should be average level for most "typical" players)... after that, a huge progression slowdown leading up to 10. This way, the casual player can feel like he's making an effort. He's contributing, he's advancing... and doesn't have "newbie" syndrome because he's the only 1st level guy surrounded by a bunch of 5th level characters... but at the same time, the non-casual players get a big reward for making that extra effort to make it to 8th or 9th and higher. Short scripted quests, quick one-off DM quests. Jump in, have fun, jump out.
CoA is NOT a professional MMO. We don't have micro-transactions to pay developers and DMs for their efforts. There's no Pay-to-Win or cosmetic "game Barbie" mechanics... While it is a popular suggestion, I think player-customizable housing and player-led factions (with short-term exceptions for specific plots) is wasted effort. There's already so much to do, prioritizing cosmetics is only going to cause more problems when the "big ticket" items don't get enough attention... not to mention that isolating PCs in their own private homes is only going to SEPARATE players from each other, not bring them together. Having a private house, while nice, isn't going to keep me coming back to a server--DM attention and good plots will. Not having a customizable house for my PC isn't going to make me leave a good server--no new quests or player interaction will.
PvP supports itself, don't need DMs for that. PvP is great for big US vs THEM mass combat war games... In D&D, for the most part, our PCs are heroes. Even if we're not good... or decidedly evil PCs... we're all playing the role of a hero, or a reluctant hero, or an antihero... PvP will happen on it's own anyway as character interaction and divergent plots intersect. There's no need to spend extra time, energy and effort focusing so much on PvP, when that time could be spent doing other things which will have a more direct benefit to the server as a whole.
Less forum, more Game. If I only have an hour to play today... I don't want to spend 20 minutes of my limited time on the forums getting updated on the most recent rumors and plot advancements since the last time I logged in. While a great place go AFTER I'm done playing, or in between play sessions, anything that happens in game should be in-game. You should be able to get rid of the forum completely and not miss anything. Rumours, Scrawls and Word of Mouth in Old Town, Announcements from the City, Declarations from the City... They should all be in-game, awaiting the daily reset, shortly after, if not before, they get posted in the forum.
More DMs. As mentioned above: easier said than done... but beyond advertising, DM attention & interaction is what will keep players coming back day after day after day. You can't do that without enough DMs to provide 24-hour coverage. In the beginning, it'll be hard... you might even have more DMs than players from time to time. But as player growth and retention takes hold, it will balance itself out. Just as in business... you have to invest into your company and risk taking a short-term loss in order to build for big long-term gains. Invest in CoA--hire more DMs. 3rd Edition D&D was built around a 4-PC party with a DM... that translates into 20% of your entire active player-base being DMs at any given time. Not only that, with the changes and suggestions mentioned above, by myself and others, you'll need that many... when they're not running quests, they can be building areas, designing new quests, creating named NPCs, writing adventure hooks and rumors for unnamed NPCs to give out. What's your target goal? 25 players online at the same time? Then have 5 DMs ready to take care of that level of activity before you get it. If you don't, someone will be left out and won't return. Prepare for what you want to have tomorrow--not what you had yesterday... if you don't, your virtual "sales" will never increase. And, business analogies aside... the players are CoA's "customers". Cater to your customer and you will attract more customers.