Loren Eiseley, the estimable archaelogist, writer, philosopher, scientist (the list can go on and on — he was a seriously smart and interesting dude), once wrote, about humans and the evolution of consciousness:
We are process, not reality, for reality is an illusion of the daylight — the light of our particular day.
And I am going to do something that only a marketer would do, I am going to appropriate a magnificent and deep thought and apply it to a marketing topic.
Attribution is a process, not reality, for reality is an illusion of reporting — the reporting of our particular day.
I talk to retailers and marketers all day — and there are two big things that they want to know:
- What Did Happen
- What’s Going To Happen
In understanding what did happen, marketers want to learn about what happened so they can be better prepared for what is going to happened. As my mother used to say, “Forewarned is forearmed.” And all marketers want to break down the past to simple cause and effect: I performed this action and got this result. Sadly, customers aren’t so interested in giving marketers direct causality — they have their own ideas.
Enter The Customer Journey
Customers come to you in all different ways — some by search, others through Facebook, others direct URL entry. The list goes on and on. If you have 10,000 transactions per month, you have 10,000 journeys to your checkout button. Attribution is supposed to normalize all your customer journeys so that you can make decisions. Attribution could tell you that your customers interact with 3 different channels before they make a purchase, and that Channel A contributes X% to your sucess, Channel B contributes Y% of the sucess and Channel C the remainder. And, then maybe, it could tell you something like Spending in Channel B will help you grow faster.
So Buckle Up Kids — Here is the Straight Skinny
Attribution is a story. It’s a fable. It is an anecdote built on a set of facts.
That doesn’t mean it is wrong, it just means it isn’t THE TRUTH. And in the world of direct causality that marketers want, that is hugely threatening. If my attribution isn’t true, then it must be wrong! That isn’t it at all. Attribution is a story from a particular point of view, e.g., your attribution model. (Here is what Google says about attribution — it isn’t expansive, but it is clear.) But here is another shocker — you need multiple attribution models — because looking at just one model might lead you down the primrose path (that is another thing my mother used to say). Reality, as Eiseley says above, is just the way we look in the light of this particular day. So other attribution models are just as valid, and may have a point of view that is crucially important in helping you make a decision.
For example, Suzy clicks on a paid search ad, and buys something. A day later, she comes back via URL direct and buys something else. How do you attribute those purchases? You have a few options:
- Purchase 1 gets attributed to the search click. Purchase 2 gets attributed to “Direct”. (This is last touch attribution.)
- Purchase 1 gets attributed to paid search and Purchase 2 gets attributed to paid search because it drove the engagement that started the purchase activity. (This is commonl known as “All Touch” or “Last Channel Touch”)
- Purchase 1 gets 100% attributed to paid search and Purchase 2 gets partially attributed to paid search, because paid search drove the engagement related to the purchase, but may have not been the driving force behind Purchase 2. (There are various ways to think about this relative to time or position in the activity chain prior to the purchase.)
So, which approach is right?
THEY ALL ARE.
Looking at transactions in the “All Touch” method can help you identify keywords that drive compelled consumed who make multiple transactions. Looking at the transactions seperately (last touch) can help you identify the impact of your brand and consumer experience on the consumer (in this case, the impact would be high because they liked you enough to come back a second time.) Partial attribution is great because you can credit the channel that started the engagement (paid search) while understanding that brand or experience may have had something to do with purchase #2.
It is really crucial to understand that attribution modeling is a reflection of the truth. You can’t see what is really true because all you see is a clickstream, you don’t know the customer journey, and without knowing the mind, circumstance, and competing opportunities in front of the consumer, you never get to understand what caused the transaction.
But If Isn’t Right, How Can I Use It?
Attribution is critical for today’s marketer because of the myriad channels and influences that impact consumers behavior. Marketers need to understand that there likely is no direct causality between your actions and any individual transaction. However, using multiple forms of attribution modeling, looking at multiple refections in the light of today, you can create a model that provides a proxy and captures the overall impact of your efforts and spend. When you compare and contrast those shimmering images against the other model’s refections, you can find enough truth to guide your actions. You need to create a process that looks at disparate versions of the truth as all equally valid in order to create a picture that can drive the best set of actions to get you closer to your intended goal. (And that is just about the suckiest marketing advice that I have ever doled out because it doesn’t tell you what to do, but rather what to ponder — and marketers, being the action-oriented type, want to do, not ponder.)
This may all sound like it is ephemeral equivication that ignores the fact that you are spending money to acquire visitor (cause) to drive purchases (effect). This is a complex world, and simplicity is to be cherished, but there is no simple answer. Marketers need to embrace the complexity of channel attribution and wield it to accomplish myriad goals. Using all touch can help you optimize for revenue. Using last touch can help to define brand impact and preferred channels. Time decay might help you understand which channels are prevalent in the awareness phase relative to the engagement phase of purchase. (Oh, and let’s not even throw in the concept of variable look back windows — I don’t want to make your head explode — but we will talk about that, “in its own time”, as my mother would say).
Embrace complexity. Find ways to collect and manipulate data so that you can capture as many reflections and illusions to drive a process of attribution that helps you grow your business. It will be hard, and it will generate two equally valid points of view that drive you towards seperate actions. Embrace the process — it will make you grow (as a person and a business)!
This post originally appeared on the AdChemix blog (a company I founded & since acquired by EliteSEM – a great place for incredible SEM/O, etc.) All opinions mine.