For somebody dedicated to finding truth in data, I sure do enjoy a good lie. Not lying in a way to try and get away with something, but more so embellishing the truth to make something more ridiculous for a laugh. Such as acting like I can’t read in a meeting, or pretending to forget something important my wife told me.
I have a friend who hates magicians for the same reason I like them: magicians are liars. They spend their time saying they can pull cards to the top of the deck or make people disappear when really it’s just sleight of hand or a gimmick bought in a store. And just as I like magicians and my friend hates them, information is also binary; it is either truth or a lie.
This makes the collection of information all the more important, because any doubt about a data point throws its legitimacy into question. But in a marketing world of thousands of platforms and millions of places where people can spend their time online, how can we say definitively that one tactic or channel leading activity before a sale is the reason why the sale happened? When determining ROI, what is truth and what is sleight of hand?
Attribution models are systems to assign value to sales from marketing tactics, based on tracked activity. I’m going to walk you through the three types of models out there that can be set up for your marketing. Like most martech stacks and systems, these are customizable based on your business and purchase process (surprise surprise, amma right y’all?), so one method might ring true for a business, while the very same could be an outright lie for another. I guess it’s up to you to see the magic and decide for yourself?
First-touch attribution
First-touch correlates revenue to the first known generator of traffic or a lead that results in a dedicated purchase. So, if I’m selling shoes…
- A person clicks on a display ad that leads to my site,
- Bounces after exploring the product,
- One day later views a retargeting ad on Facebook but doesn’t click it,
- Two days later, after a Google search for our site, comes back to the product,
- And ultimately purchases.
Using a first-touch model, the sale would be attributed to the ad that first drove the visitor to the website. The rationale here is that the first ad began the process of interest for the prospect.
Last-touch attribution
Last-touch correlates revenue to the last known generator of traffic or a lead that results in a dedicated purchase. So, back to selling shoes…
- A person clicks on a display ad leading to my site,
- Bounces after exploring the product,
- One day later views a retargeting ad on Facebook but doesn’t click it,
- Two days later, after a Google search for our site, comes back to the product,
- And ultimately purchases.
Using a last-touch model, the sale would be attributed to the Google search (or SEO) that last got the visitor to the website right before purchase. The rationale here is that the Google search was the final piece of the puzzle and got them on site to finally buy.
Multi-touch attribution
Multi-touch correlates revenue across all known generators of traffic or a lead that result in a dedicated purchase. Last time with the shoes…
- A person clicks on a display ad that leads to my site,
- Bounces after exploring the product,
- One day later views a retargeting ad on Facebook but doesn’t click it,
- Two days later, after a Google search for our site, comes back to the product,
- And ultimately purchases.
Using a multi-touch model, the sale would be attributed to and divided in some fashion across all the tactics engaged by the user in the known lifetime before purchase. The rationale here is that everything the user saw, from the first ad to the landing page and even the Facebook ad that they just looked at, played a part in reinforcing the need for a purchase.
Depending on how your business operates, with a custom multi-touch attribution model, greater values could be assigned based on timing or engagement. So for a shorter time to purchase, maybe we give more credit to the last touch tactics than the first (First: 25% of revenue, Mid: 35% of revenue, Last: 40% of revenue). For a longer purchase cycle, it would likely be flipped, as the generation of the lead to allow for longer term marketing is more critical (First: 50% of revenue, Mid: 25% of revenue, Last: 25% of revenue).
How to escape an attribution straitjacket
To deploy a marketing attribution system, you’ll need any number of software packages to collect and analyze user activity across your utilized marketing platforms. Some platforms are totally dedicated to tracking attribution as well, or you can employ a more manual method to divide sales revenue across any number of touch points.
We’ve made a handy tool if you’re in the market for a new software solution, check it out for reviews and pricing information broken out by SaaS category. First, be sure to think strategically about your attribution model and how it plays into your stack. And of course, Element Three is here if you need help in that realm.
Apparently, Houdini would pop his shoulder out of place to actually undo his straitjacket, which is a crazy personal feat that seems like magic, but really is true. So with the shades of gray all around marketing activity, it’s important to strategically decide what you will consider to be true, and move to collect and improve those metrics.
