Well, you think, Jay, for many years, we said that we felt the seat opportunity in a given account for Windchill was 10x to 20x the seat opportunity in that same account for CAD. And so we're headed there. As you mentioned, we're at sort of a 7x ratio right now. But that said, we're selling windchill to a lot of accounts that never bought any CAD, so you can't just take our CAD based on 7 and say that's the opportunity because when you think of some of these big retail accounts, the targets, for example, they don't have any Creo and we're selling them many, many Windchill seats and we've had a terrific run of those type of accounts. And then on the other hand, we're selling a lot of Windchill to accounts that use somebody else's CAD tools, use CATIA or UG NX [ph] or what have you -- SolidWorks. So I think that we feel like this ratio will continue to develop in the direction you said. Now I think there's also an opportunity to increase the CAD seats in this cross-sell of direct to Parametric and Parametric to direct technologies should as well help with the cross-sell and could drive up the number of CAD seats, thereby further increasing the number of Windchill seats. So I think from a seat count standpoint, we expect this trend to continue indefinitely. The second part of your question, I think, was really about the ASP, the average seat price. I think that there is 2 competing trends there. On one end, as the footprint of Windchill gets bigger and deeper and we do have more modules that we can pile into an existing seat and then the price of that seat and that user goes up. But on the other hand, we keep holding in more and more casual users as we expand to that bigger ratio. As you go from 7 to 1, to 10 to 1, to 20 to 1, you're involving more and more casual users, who would expect to pay much less per seat than sort of the more concentrated users that preceded them. So I think these 2 factors balance each other out, and I can't really predict that the average maintenance per seat will go up. I think the 2 factors might kind of counterbalance each other, and we would see about the same trend, if I had to guess.