I’ve spent the past two months trying to figure out what it means to be PDC Content Owner, and how I’m supposed to do my job.  I’m not entirely sure I’ve figured it out yet ;)  I know that by Sep 13, I need a plan for how to keep several thousands of attendees excited by, and interested in, a couple hundred breakout sessions, hands on labs, panels, symposia, ask the experts sessions, and more.  There’s also the huge question of how we foster an excellent sense of community at PDC, but fortunately there is another team member who’s leading up that charge (he hasn’t blogged about it yet, so I won’t be the one to blow his cover ;)

The raw materials to pull all this together:

  • The plan from last 2003’s content owner.  Steve’s working document for his PDC planning last year was a 100+ slide PowerPoint deck, where he’d keep track of all the things he was working on.  Sort of like a PPT blog, I guess, only out of date content gets deleted.  It’s an amazing reference point, and if you read it right, it becomes a “how to develop content for PDC” manual.
  • The raw data from PDC 2003.  There are three particularly interesting buckets of data from the 2003 event.  First is the pre- and post- event survey.  We did a phone survey of PDC attendees before and after the event, to see why they were attending PDC and what they thought about the event.  It provides some good data, I’ll see if I can blog about it in the future (not sure about confidentiality of the data.)  Second we have the session feedback forms, which we asked attendees to fill out after every session.  Those scores help us understand what content attendees found most interesting/valuable (although actually it’s hard to distinguish between those two sometimes.)  Third, we have the attendance count for each session.  Given that the top complaint in session feedback forms was that many of the sessions were overcrowded, the attendance counts help us understand how to do better for 2005.
  • Microsoft’s product teams.  I have been making time to speak with architects and senior product leaders from across the company to get the latest updates how they’re progressing towards the vision we talked about at the last PDC.
  • The Los Angeles Convention Center.  It’s pretty daunting when someone gives you a map of LACC and says “okay, tell me how you want to configure all these rooms for your sessions.”  Fortunately, we have the layout/plan from 2003 to start from.  We have a great events management team working on PDC, and they’ve juggled stuff around from 2003 so that we opened up several additional rooms for breakout content.  That is good news, since it potentially means we can provide more content.  On the other hand, several of these new rooms only seat ~250 people, which as you'll see below is a  bit inconvenient.

 

Given all this data, I’m trying to come up with a plan for both the what content we cover (which products/technologies need the most breakout sessions/labs/etc), and how we cover it (how many sessions can we actually schedule, how many repeats do we need, etc.)  Given how strong the feedback was on session capacity in 2003, one of things I’ve focused on in this early planning stage is how to improve the way we match sessions with room capacity.

 

The data from 2003 show that nearly half of our sessions had between 200 and 400 attendees.  Another 25% or so had under 200 people.  This is particularly vexing, because it turns out the LACC has a lot of rooms which, after we put in staging and A/V equipment, can seat just under 300 people.  You might think that the easy solution is to take those 25% of sessions with <200 people and put them in the <300-person rooms.  And yes, given the data we have now about 2003 actual attendance, we could easily schedule every 2003 session in a room that holds everyone who wants to attend.

 

Of course, my job isn’t to reschedule the 2003 PDC in retrospect, it’s to schedule the PDC 2005 with foresight.  And it’s so early right now that I don’t even have a list of sessions – we’re still trying to just figure out how many rooms we should use, how many total sessions we should allow for, etc.  One of the things that makes this so difficult is that our ability to predict attendance at a session is quite poor.  At about 6 weeks before the event, we ask attendees to build out their likely agenda using the CommNet scheduling tool (which provides the entire session list and scheduling information,) and we try to use that to model attendance.  In 2003, it was not a very effective modeling tool.  We had only a small percentage of attendees who chose to fill out their complete schedule, and it turns out the preferences of those folks did not accurately reflect what the majority of our attendees would want to attend.  Looking at the 2003 seat counts, we managed to get about half of our sessions into rooms that held more or less the right number of people.  Which means of course that half the sessions were way off base.  The good news, I suppose, is that in most cases, we erred on the side of caution – scheduling a talk with far higher capacity than the actual number of attendees.  But we also had about 10% of our sessions where we were off by upwards of 50%, like when a talk in a 280-person room draws 400+ attendees trying to get in.

 

So if we know a lot talks will have between 200 and 400 attendees, and we know that a fair number of LACC rooms can’t hold above 280 people, and we also know that our ability to predict which sessions will draw more than 280 people is poor, how do we approach our capacity planning?  We’re in the midst of figuring that out now, and hopefully I can blog about some of the solutions we come up with.