My business partner Mike Baird is a UX genius. He drives me crazy. While I’m trying desperately to build a product that works in beta, he’ll say something nutty like this: “Instead of focusing on what our product needs the user to do, let’s talk about what the user wants to do with the product.” That’s actually a direct quote from two nights ago.
He was talking about the step where the user defines the pattern for their new codes. Admittedly, it’s the most confusing piece of the puzzle, because marketers don’t want to think about how the tracking codes are formed, they just want a new tracking code. Just like I don’t want to know the etymology of the word “cheesecake”; I just want cheesecake. Specifically, I want a slice of that turtle cheesecake they have down the street at Kneaders Café. Are they still open at 10pm on a Thursday night?
But if I have a tool that’s going to generate new codes for you, there has to be a method in it. Although some people use randomly-generated codes for every initiative, which you can actually get away with if you describe each initiative clearly and completely in SAINT, I don’t recommend that approach. I like codes that you can decipher later on if you need to.
But more importantly, the decision of how to form your new tracking codes is made by individual companies, and they each do it their own way. I want to respect that. I want my tool to let people keep doing it the way that they’ve decided is right for their business.
But that leaves me in a UX quandary: how can I make cheesecakes for people using their own home-made recipes, without them having to know or even think about the recipe? And the answer is the Pattern Wizard.
The data that Omniture stores, after all, isn’t just a list of tracking codes, but lots of columns defining what those codes mean–and while the tracking codes themselves may be cryptic, those definitions are written in the language that marketers already speak.
So right off the bat, we ask the marketer question number one. I actually can’t tell you what that is, but I’ll know it when I see your SAINT file. If, for example, you have a classification set up called Marketing Channel, where 90% of the time the value is either PPC, Social Media, or Email, then question number one would probably be:
What Marketing Channel are you creating codes for today?
That’s a softball that any marketer can hit, and the coolest thing is that the answer allows us to devise the next question. Because if they say PPC, then I can discard everything in the SAINT file that’s not PPC. I may see, for example, that when the Marketing Channel is PPC, then the classification column Publisher is almost always Google or Bing. And so on until all of my questions are answered.
The end-game of the wizard is not only to simplify the table the marketer will fill out on the Classify step, but also to predict what the typical syntax of such a code will look like. We’re using the plain-text SAINT file to reverse-engineer the cryptic elements of the tracking code, producing a recipe that the marketer needs only to accept after answering questions that we’re sure they know the answers to.
Of course, our wizardly questions will rely on the quality of the SAINT file that we receive. But even if your company’s SAINT classifications are poorly defined (or currently unused), the beauty of our system is that, each time you use Claravine, your SAINT files will become better and more predictive for when you return.
It’s a solution that even Mike approves of–and after years of collaboration, that in itself is no small victory.