Attribution. It’s one of the 6 Ideas that’s driven UX conversations this year and let’s just just get this out of the way right now.
I thought it was total crap. Something made up by bored marketers to sell big business that we can do now. Because, you know, data.
But I was way wrong. Attribution makes a lot of sense for certain marketers. For others, it’s overkill.
If you’ve heard the term before but aren’t quite sure what it is, this post is for you.
Attribution: What We Think It Is
When I first heard about marketing attribution, it sounded like the holy grail of advertising. The promise was that we’d know which ad caused somebody to make a purchase.
But then you think about it for 2 seconds and come to a simple realization: that’s impossible.
Even given perfect data, how are we supposed to know why somebody decided to buy something? Doesn’t it give the ad too much credit? I mean, if I were in the market for a new car, how many Ford ads have I seen in my lifetime and which one got me to show up at a dealership? Isn’t it more likely that I have a personal situation that makes me need a truck and in my research and in talking to my buddies, Ford’s F-150 seemed like the answer? How does a particular ad play into my buying decision? And more to the point: so what!
Isn’t the whole point of the exercise to realize that buying decisions are complex and can’t be reduced to a particular ad? Oh yes! It is!
So… right. That obviously sounds stupid.
And nobody preaching the Word of Attribution is telling that story. It’s just easy take the word ‘attribution’ literally and run with it.
Attribution, in reality, is easy to understand and makes a lot of sense. In fact, you’ve already been doing it. You just haven’t taken it to its logical, data mining, extreme.
Attribution As You’ve Already Done It
Before we really look at what attribution is, let’s look at how you’ve done already probably done it.
First, you already use Google Analytics, right? Good. And you’ve defined goals and have assigned values to them, right? See? I knew you were good.
When you look at your analytics profile, you see conversion data. You know which keywords are generating the most revenue, which referral links are money makers, and which ads you buy are generating sales. You feel like you’re pretty ahead of the game when you talk to your friends. And you are.
But when you see those dollar values next to the links, or keywords or what-have-you in Google Analytics, what you’re seeing is last click attribution. Google Analytics is telling you that the keyword, or a link on a referral site, or a particular AdWords ad is the last thing that users clicked on before they came to your site and made a purchase.
According to Google, marketers find this wholly inadequate. “Only 14% of respondents consider last-click analysis (which, until recently, has been the industry standard) to be “very effective.” Yet over 50% of them are still using last click measurement – most likely because they haven’t yet found or mastered the right tools to take them beyond the last click.”
Google brings up a good point about tools. But before we get there, let’s look more closely at what Marketing Attribution really is.
Marketing Attribution Is…
As we intuitively know – and griped about above – purchasing decisions are complex. Even inexpensive products get discovered, researched, compared and purchased across multiple channels and touch-points online. Basically, users spend time with a product before they buy it. And this time is spread across all the possible places they can interact with what you sell online.
What marketing attribution does is provide insight on which channels and touch points have the most influence in informing the buyer.
Or, to take the marketing-speak out of it… Marketing attribution assumes that marketing is a team game, like football. As the team drives down the field, certain players make key plays. When a player scores, he gets the credit. And in last click attribution, we’d just say the guy who scored was responsible for scoring. What marketing attribution really wants to know is who else on the team made plays that contributed to the score. Only, they want to know that over time. So instead of looking at an individual game, look at a whole season. Do certain receivers play better when playing for specific coaches or quarterbacks? How do quarterbacks play when different players are inserted on the offensive line? How does the defense affect scoring? That sort of thing.
Think of it like fantasy football, or designing your Career team in Madden. You’re looking for players who are ballers in their own right but who are also a good fit with the other players. If you have players who aren’t performing, you want to be able to see that and either fix it or replace them.
In marketing, it’s the same thing.
Let’s take a very simple example:
You have a website and somebody finds you through Google’s organic search results and makes a purchase.
You also run banner ads online.
Marketing attribution wants to know if users who see your banner ad AND find you through Google organically are more likely, less likely, or as likely to buy as users who ONLY found you through Google or ONLY saw (and clicked on) the banner ad.
In essence, are these two mediums (AdWords ads and organic search results) good teammates? When they play together are they more likely to score than when they play separately (or with others)?
Now, in real life, you have many different “players” to consider in search, social, referral, email, banner ads, and more.
Okay, you’ve sold me on attribution. How do I do it?
According to the same Google white paper mentioned above, companies use a range of attribution methods.
It seems clear that there are a few different ways to go about it:
– Last click
– First click
– Custom weight
– Linear weight
– Time decay
– Voodoo (possibly)
Let’s blow through what these mean, although the names are pretty self explanatory.
Last click – The last thing a user clicked before landing on your site.
First click – The first impression registered by that same user before they ultimately made their purchase.
Custom weight – Something determined by the client/agency/company that is specific for their needs.
Linear weight – All registered hits are counted equally.
Time decay – The longer back somebody interacted with a touch point, the less it counts
Voodoo – Black magic.
Now, if you’ve just read through that list and went, “Wait, can’t we use a combination of first, last, linear, time decay AND voodoo?” And, it’s possible. I’m going to recommend leaving out the voodoo, but certainly, it’s worth using all of these methods to extract the most insight possible.
The Biggest Problem with Marketing Attribution
In a word, politics.
Marketing attribution is still the provenance of Big Business (and to a lesser extent, mid-size businesses), so marketing attribution is going to be political. Attribution is going to affect departmental budgets. So, it’s a great too. It’s not smoke-and-mirrors. But it’s a double-edged sword. It makes marketing make more sense. BUT, it also cuts under-performing segments. Or at least you think it would. But according to Google, the main affect of marketing attribution is to increase spend in the places that work. I’m guessing the reason the budgets don’t go down in other areas is because of the conservatism towards believing the data being completely reliable. It’s not as politically negative to reward performing segments while not cutting under-performing channels.
But really, I’m just speculating.
If you work for yourself, then it’s all upside. 🙂
The Technology of Marketing Attribution
Marketing Attribution is still in its infancy. And the tools reflect it. Nevermind the Adobe commercials, most people are doing something considerably more low-tech.
As you can see in the following chart, of the four options, the number one way people are doing marketing attribution is with Excel spreadsheets.
Dude, that’s EARLY. That’s like, “Oh, look at me! I’m a marketing hacker! I use Excel!”
Can you imagine where these tools will be in a few years? I know one thing, they won’t cost $35,000+ a year and be sold to Fortune 500 companies.
It’s hard for Google to conceal this fact in the report. They might as well say “most everybody is still making it up as they go along”.
However, many client-side organizations also use multiple techniques and technologies. Among marketers, experience seems to be a key variable in sophistication. There’s a correlation between the number of years an organization has been practicing attribution and the likelihood that they use multiple inputs and approaches.
This can only mean one thing: experimentation still rules the day.
Marketing attribution is a promising development in data mining, one that helps refine how to reach your most likely-to-buy customers. At its best, it suggests relationships between marketing channels that are effective at driving sales. Exactly how to do that, well, there’s a lot of developing theory on that.
In next week’s article, we’ll look at some specific instances of marketing attribution and show what the spreadsheets look like. We’ll look at what we can learn with last click, first click, linear weight, and time decay attribution. I’ll see you then, to close out 2012, on Dec. 31st.