'Third Party Dependencies', or 'Have an Backup Plan'

The interconnectedness of the web is a fantastic thing. There are tons of libraries, expertise, content sources, services, etc out there that make it possible for companies to focus on their core competency. If you are a social networking site, do you really want to build a video player, for example? Maybe, but only if that video player is part of the strategy that is your core competency, or that gives you some sort of competitive advantage. If Previously, you would have to serve your own photos, but maybe now you can use a service like Photobucket, or Flickr. Let the photo storage be their problem - let them handle the disk space requirements and the server load of serving endless quantities of images, right? Right?

Well the answer in my view is maybe. Let's face it. Sites like photobucket aren't going anywhere, but their terms of service might. What they are allowed to do with your content might change, and the conditions of your agreement just might become unfavorable, and there is nothing you can do about it, because now you have given all your content to them, and you don't really have any of it. This could be a bad thing, right? Especially if you aren't able to at least get a copy of it back from them without going through some manual pain.

Many companies simply won't make it. You probably don't want to dive into partnerships or use services like this from companies that you aren't sure are established, because chances are, you will have to start over sometime, and it will probably be really inconvenient. Then your stuff might just be gone, stuck on a server somewhere that gets sold at auction. Prognosis: NOT SWEET.

My suggestions on picking a partner (based on some recent experience) would be this:

1) Find a partner who can solve the business need, first of all.
2) Find a partner that isn't going anywhere (requires crystal ball in some cases)
3) Find a partner that will let you get your 'stuff' back

Once you have chosen a partner, come up with a backup strategy. You are sending content to them, but can you periodically backup this content to a disk that you own and control? At least this way you have the files if something goes wrong, and you can then bulk load them into another tool if need be. Then you are limited to the code changes you have to make to do the API communication, and this should be relatively limited. Rarely do people get things like this right on the first try, and frequently they are lucky, because they never figure that out - sometimes you just try something and it happens to work, but many times it doesn't. If it doesn't, remember the 'fool me once, fool me twice' adage.