Yeah, Crispin's got the basic core process down right. Just tracking your keywords by subid is wasting a ton of information. What you should do is assign each clickthrough a unique transaction id and then attach all other pertinent information to that transaction record. That way you can go back later and true up the revenue data with the exact transaction instead of just with a general keyword. But since you also save the keyword (and the network id, campaign id, adrgoup id and whatever else) you can tie everything back together and get a lot more data out. There's a lot more you can do, but that's the basics.
Generally speaking, people get too hung up on what networks do for them by default - "subids" and "google analytics" and the like which are imperfect and by definition outside of your control - because they either don't have the time to build the internal technology, don't understand the value of the internal technology or are too small to warrant building the internal technology to support their marketing efforts. These are all perfectly valid reasons for using a random cobbling together of disparate, aftermarket tools. But if you're going to have a ton of transactions and you need your own accounting systems or business intelligence systems then, instead of spending your days copying and pasting reports from five different networks into Excel and doing data sorts, you should begin by thinking not about what data you must gather, but the entirety of what data it is possible to gather and how to gather it (and believe me there's plenty of data that isn't available by default). That way, once you figure out later that there's another two data points, you can combine to figure out a useful metric (and you will) you'll already have the data in your database. Then you can begin cross-channel targeting and all kinds of other neat things.
If you plan to record things, then it's a minimal investment up front to record all kinds of stuff, you just have to figure out how to do it and what's available. And, no, no one thing (or five things) off the shelf can do all this for you, though if you just want to work between one common channel and another common channel, there are OS or off the shelf apps out there to help with that. As soon as you want to add a third channel, though, you're generally back to square one.
Generally speaking, people get too hung up on what networks do for them by default - "subids" and "google analytics" and the like which are imperfect and by definition outside of your control - because they either don't have the time to build the internal technology, don't understand the value of the internal technology or are too small to warrant building the internal technology to support their marketing efforts. These are all perfectly valid reasons for using a random cobbling together of disparate, aftermarket tools. But if you're going to have a ton of transactions and you need your own accounting systems or business intelligence systems then, instead of spending your days copying and pasting reports from five different networks into Excel and doing data sorts, you should begin by thinking not about what data you must gather, but the entirety of what data it is possible to gather and how to gather it (and believe me there's plenty of data that isn't available by default). That way, once you figure out later that there's another two data points, you can combine to figure out a useful metric (and you will) you'll already have the data in your database. Then you can begin cross-channel targeting and all kinds of other neat things.
If you plan to record things, then it's a minimal investment up front to record all kinds of stuff, you just have to figure out how to do it and what's available. And, no, no one thing (or five things) off the shelf can do all this for you, though if you just want to work between one common channel and another common channel, there are OS or off the shelf apps out there to help with that. As soon as you want to add a third channel, though, you're generally back to square one.