In this series of articles I’m going to discuss the basics of multichannel attribution and how you can go about understanding the contribution that your marketing brings to your digital conversions.
This is an area of marketing which often slips somewhere in the cracks between the IT side of the business, the business analytics side of the business, and the marketing department. It’s actually not rocket science, and this series of articles will take you through the basics.
In this first article I’m going to give you the 100 mile overview of digital analytics. I’ve broken digital attribution down into 5 steps which I’ll introduce here. Each area will then be explored in more detail over the coming weeks.
To understand what’s driving traffic to your website you have to start by classifying traffic, particularly paid traffic, so you can understand what’s driving your conversions. Are they being driven by natural traffic, paid google search, display traffic, social networks, etc. etc.? Some of this can be done by looking at the referrer, but it’s also good practice to identify the incoming traffic by marketing channel. This can be done simply by adding a unique identifier to all of your paid advertising target URLs.
For example, if my website is woo4.me, I can add a query string to the URL on my paid-for advertising to identify the source of traffic. e.g. woo4.me?ch=1&ca=7 tells me that traffic used channel 1, and campaign 7. The channel attribute would be different for each marketing channel I used to target traffic to my website. eg Google paid ads would have channel 1, Footer links for my social profile would have channel 2, paid links from twitter would have channel 3.
Think about the marketing channels you use for your own website and separate these into your own channel list. Now make sure you use them correctly!
Once you’ve got a classification for incoming links, you can now identify the key marketing events which have occurred during a visitors interactions with your website.
We’ll go into this in more detail in the actual article, but the goal of pathing is to get to something like this…
PPC -> SEO -> SEO -> DIRECT -> *CONVERSION*
Every single visitor to your website will have a “path” but for most attribution tasks we only really care about those that lead to conversions. (of course there’s an exception to this rule, but we’ll get to that in the detailed article for section 3…)
Once we’ve got the path associated with a conversion we now need to attribute that conversion to a marketing channel. Let’s take the example path we had above, there are a few basic attribution models that are typically used to allocate the conversion.
The one you’re most likely to encounter is “last-click”, this gives all the credit to the marketing channel touched immediately prior to the conversion event. In this case that would be “DIRECT”.
A “first-click” model would instead take the introducing channel, in this case “PPC” and assign the conversion to that one instead.
Another popular model that marketing teams like to use is the “any-touch” model. In this case the PPC channel, the SEO channel and the DIRECT channel all get credited with a conversion. You can see why the PPC team, and the SEO team would like this model. But if you sum up all of the claimed conversions you’ll generally end up with more conversions than actually happened.
There’s a number of reasons companies end up with “any-touch” attribution, and it’s mostly accidental, but we’ll talk about that in a later article, along with more complex attribution models like linear, Bayesian, etc.
Now that we have a bunch of “conversions” assigned to each marketing channel we need to work out what the value of each marketing channel has been. In a simple world we start by working out the total cost of activities that support a particular marketing channel. We then divide this number by the total number of conversions attributed to the channel. This gives us the CPA (cost per acquisition) allowing us to do simple comparisons between different marketing channels.
However, this model assumes that all conversions are equal, and it’s never that simple. Let’s assume that I’m an ecommerce vendor. Getting people to pay for an item might be my initial conversion. But I really care about the value of that conversion in monetary terms. Was the purchase high value or low value, did the customer go on to buy more products later?
With these more complex conversion attributes I can now revalue my marketing channels and find out which channel brought the most monetary value instead of purchase events. If I measure each conversion by the income generated, and then attribute that to the marketing channels, I’d almost certainly discover that while some channels are good at generating low value one-off purchases, other channels are better at generating recurring high value purchases.
Finally I need to optimise my marketing activities. If I don’t do this then all I’ve really achieved is a good understanding of how effective my marketing is. The business only benefits if I can then use this understanding to improve my marketing. In the final article I’ll discuss ways to do this for general marketing channels, although your marketing channels may require different methods!
I hope you’ve enjoyed this brief introduction to digital marketing attribution, and join me again in a few days for the detailed article on channel classification.