Third-party add-ons are available for most services and features not natively supplied in core, and these are usually well-tested and well-supported
Drupal development brings a series of coding standards that would make our internal collaboration considerably easier, as we'd adhere to those standards
Modules are available which can be used by developers to instill a more disciplined development process: there are essentially plug-and-play modules for PHPUnit, Behat, and Code-Coverage tests, among others. We could take this opportunity to have a VERY disciplined development process.
Getting Drupal coded to the point where we already are with EE would be a significant lift
Large installed base means that security holes are regularly found and exploited (but are fixed relatively quickly)
Community is made-up of a very inconsistent userbase with wildly-inconsistent skill levels. Using third-party add-ons needs to be done carefully, to avoid implementing the code of maniacs
There is a learning-curve. For most of what we'd need to do, it's not terribly steep, however.
Pros of running our own instance of Drupal, versus a hosted solution:
Unfettered level of control. We could do just about anything that is possible with Drupal, which is just about anything web-based. We wouldn't ever have to tell Marketing (or others) that we "can't" do something -- we could partner with them to determine when and how we could do it, with a clear idea of how heavy the lift is, how long it would take, and ultimately what it would cost.
Responsiveness of the system and development processes is totally under our control
We have people who can run, code and tune it already on-staff
Enhancements requested by business units can be prioritized by us, rather than negotiating with the host for them.
I have a pretty nice starter-build of Drupal that could give us a boost in bootstrapping an AKC Drupal.
Customizations made for one AKC Drupal instance could be ramped-up for any additional AKC Drupal instances at our discretion without additional contracts and restrictions (example: we could share akc.org improvements with akcwinners.com, etc, at no extra cost).
Potential to serve the dog community: it would be very possible to release Drupal installation profiles keyed for the needs of various segments of our enduser base -- for instance, vets, breeders, trainers, etc. Once started, it would be pretty easy to share content with and from those endusers.
I'm reasonably sure that any hosted, proprietary solution would charge us for each ancillary site, whereas with our own custom multi-site Drupal build, we could (for the cost of labor, and possibly server resources):
Have a basic AKC theme with potentially all the features of base AKC.com, which could be altered and used on sites added via procedures we'd control (for a very basic site, it could happen in a single workday for a single developer)
Share a set of credentials across all AKC sites, giving our workers a single-sign-on to every one of these sites, with appropriate and configurable permissions.
When there are security holes discovered, we could plug those on every one of these sites at once, minimizing the work required to respond to security alerts.
Use these additional sites as additional sources of content and traffic
For commonly-used integrations with external sites, they could be used in every site in an identical way -- example: social integrations with Twitter could run identically (or as differently as we wish) with only different Twitter credentials on a per-site basis.
Cons of running our own instance of Drupal, versus a hosted solution:
More time required for setup
Assuming that a hosted solution supplies support, we'd be passing on that benefit
Presumably, a hosted solution would apply security updates and patches for us. A minor benefit, as they are almost never time-consuming.
If we decided to release installation profiles, there would almost certainly be support-requests.
Pros: