Happy 2016! Let’s kick off the new year with my favorite topic — digital analytics. More specifically, in this post we’ll talk about attribution.

During the month of December, most of you probably did some shopping (in-store or online). No doubt you saw advertising for products, whether that was through banner ads, emails, paid search, social media or other advertising. Did you take action on any of those ads or promotions? If so, in order to measure the effectiveness of each channel, that company had to consider the many touchpoints that occurred (your journey) before you took an action — before you converted or bought the product you wanted.

Introducing Attribution

Attribution has been a hot topic for a long time in the digital analytics industry, but Google brought the topic to a much broader audience when it launched multi-channel funnels and assisted conversions within Google Analytics a few years ago.

When someone mentions attribution, eyes might glaze over. But if you’re involved in marketing at your school, it’s important to understand attribution. Even if you end up using only the default model, it’s critical to understand why attribution is important. To truly understand the different models, you should test them out.

So, where should we begin? First, think about how you behave when buying a product online. Chances are you probably don’t view an ad, click on the ad, go directly to the website and buy a product immediately. Most users don’t behave this way — or rarely behave this way — especially when searching for the right college or university. However, when we use last-click attribution (most analytics tools’ default attribution model), we’re betting this is exactly how people behave.

Defining Attribution

What is attribution and what are the different models Google Analytics uses?

Let’s take the holiday shopping example from above. Understanding the many touchpoints, when the conversion took place and how we give credit to each touchpoint for that conversion is attribution.

By default, Google uses last non-direct click attribution.

Last Non-Direct Click Explained

To explain what last non-direct click attribution means, let’s look at another example:

Jack goes to his favorite technology website and while reading an article, he notices an ad and clicks on it. He goes to the website and browses around, but doesn’t buy anything. The next day Jack realizes he’d like to buy a product from that website and goes to Google to search for the website name. He clicks on the first listing, which is a paid search listing, and goes to the website. But he doesn’t buy the product during that visit. Instead Jack clicks out of the website and goes to a review website to read reviews of the product. The next day, he wants to buy the product. He remembers the URL of the website. He types in the URL, goes to the website and makes his purchase.

Here are the touchpoints along Jack’s journey to conversion (purchase):

• Jack clicks on a banner ad and goes to the website, but doesn’t convert → first touch: banner ad
• Jack searches Google for the company name, clicks on a paid search listing, goes to the website, but doesn’t convert → second touch: paid search
• Jack types in the URL directly, goes to the website and purchases a product → third (and last) touch: direct

With last-click attribution, the direct channel (which was the last touchpoint in Jack’s journey) would be 100 percent credited with the conversion.

However, in the above scenario, within all standard reports other than multi-channel funnel reporting in Google Analytics, 100 percent credit for the conversion will go topaid search. This is called last ‘non-direct’ click attribution.

Why does Google give 100 percent credit to the last non-direct click?

If you think about it, it makes sense. By definition the people who come to your site directly, that is, by the direct channel, already know you. They type in your URL from memory or they click on a bookmark they’ve created for your website. To give this channel 100 percent credit for a conversion doesn’t make sense if you truly want to see which channel is impacting that conversion.

Quick Side Note — Importance of Trackable URLs

However, we know how tricky the direct channel can be in Google Analytics (or any other digital analytics tool for that matter). What do I mean? I mean that the directchannel is the catch-all. If we don’t do a good job creating and using trackable URLs for our marketing efforts (not only paid advertising, but links within our emails, etc.), those clicks could be categorized as direct (or referral if coming from another website) when they shouldn’t be.

A quick example: let’s say you send an email to a prospect and create a link within the email that doesn’t have a trackable URL. The prospect opens that email in Outlook, clicks on the link and goes to your website. The attribution channel for that session will be direct. It should be email, but because there wasn’t a tracking URL, Google Analytics has no way to know that it’s actually coming from an email.

Attribution Models Explained

An explanation of attribution models can be found on Google’s support page forGoogle Analytics attribution models. But let’s talk about a few (I won’t go into all of them) and see if we should test them out.

We already explained last non-direct click model above.

The last click (or interaction) model is similar, only the channel type doesn’t matter. This means in the scenario I talked about above, it would give 100 percent credit to the direct channel. It doesn’t care what channel was the last touchpoint — it gives that channel 100 percent of the credit. This is the most typically used and talked about model.

The linear model gives equal credit to each channel touchpoint within the journey. If we take the above scenario, it would give approximately 33 percent credit to each channel (banner ad, paid search, direct). As Google states on their support page, this model is useful if each touchpoint is equally important.

The time decay model is interesting. It’s all about timing. It gives more credit to a touchpoint that occurs closer to the conversion. Google states that this would be good to use if you have a short consideration period. That consideration period is sometimes tricky when trying to track true conversions.

Think about this for a moment. Our consideration period is long. However, sometimes we aren’t able to measure to an application or to an enrollment (or a donation for development websites and so on) on our website. The ultimate conversion sometimes happens on another website we don’t control or manage. So if what we can measure is the inquiry form only, technically the consideration period for this conversion may be short. It might be appropriate to test this model out.

Why do I continue to say ‘test out’ instead of ‘use?’ It’s critical to have a great understanding of attribution models before using different ones to make big decisions. For right now, we can test these models out and see how they affect how the channels are credited with a conversion.

There’s your high-level introduction to attribution. In a follow up post, we’ll talk about what to consider before diving into attribution model testing and usage. We’ll also discuss lookback windows and assisted conversions.