The problem with attribution

I’m going to get a little technical here. I’m going to talk about online marketing and attribution. Firstly, as an online store, we are able to track a lot of the data from visitors who come to our site. There are 2 major ways this data is collected – logs & cookies. Log files tracking basically runs on the basis of tracking the requests a customer makes to our server. Every time you load a page, you request information from us to load it – we can track this data and theoretically assign it to individuals. With cookies, we drop a ‘cookie’; a piece of code into your browser which tells us what you are doing on our site. Cookies are also useful for minor things like keeping track of what you have in your cart, whether you are logged in and the like.

The big difference between cookies and log file tracking for the purpose of analytics is that customers can clear or refuse to take cookies at anytime. If you do that, we can’t track your data if we are just using cookies. For log files, because you have to call information from our site, we can track every visitor. The minus of course is that it’s very hard to track visitors over multiple sessions. And yes, I’m simplifying greatly. If you’re curious, Starlit Citadel do a bit of both.

As an online store, what we want to know / attribute is what kinds of marketing work for us. We want to know if advertising on Facebook makes a difference rather than say, Youtube advertising. I’ve discussed these kind of decisions before, but let’s talk about one of the biggest ‘plagues’ for us – attribution.

Traditionally, the only way to tell why a customer purchased was via their last clilck – the last place they came from. This meant that if a customer came to us straight from a bookmark and/or typed us in directly, we’d consider them a ‘direct’ sale and if they came from a Google search, we’d attribute the sale to Google. Of course, how many of you have read an article or thought of a store you wanted to visit, typed in their name in Google and gone to the store that way? I know I have, numerous times. Now, we’re attributing a sale from other forms of marketing (say a mailer sent to you) to Google because that was the ‘last click’.

In the past, we could filter some of these out by ignoring any searches that used our brand name. These days, Google has removed all that data, ‘stripping’ it in from the information that gets sent to us in the quest for privacy (really, just to make us pay more for advertising).

These days, more information is being added by tracking visitors when they leave the site. For example, Google now ‘tracks’ other interactions, keeping data and how you interacted with our other forms of media like Facebook, Twitter, etc. Again, there’s a lot of leeway here – you could see a Twitter post by Starlit but you don’t click on it, so it wouldn’t get tracked. Still, it does keep track of what you interact with, before you buy.

Now, let’s step outside of talking about our main tracking platform (Google Analytics) and talk about advertising networks. Most other advertising networks, for their own reasons, track conversions.  These conversions can vary depending on how they attribute the data and it can often not lineup with the data you get from other analytics you might have.

Advertising networks might be attributing conversions over a period of time (i.e. you clicked on an advertisement they served, so if you buy in X days, they’ll consider it a ‘win’ for them), based off impressions that you might have on the advertisements and the like.

At the end of the day, when you are looking at this information; you need to work out which data you trust, understand what the data that you are trusting is based on and then, most importantly, decide what your cut-off point is.  Are you willing to take an ROI of 3? 5? 10?