One of the biggest challenges in analytics is accurately attributing a purchase or conversion to the channel/source responsible for that conversion.
It is very rare that a conversion happens in single visit to a website. More often than not it is multiple visits that will create a conversion.
Let’s look at a simple example: Mark searches for a vacation deal in Google Search. He finds a couple of attractive blog posts that matched his keywords and organically ranked very high in Google Search. He visited three different websites, read the articles and browsed holiday deals. He then signed up to an email newsletter on each of those websites. After two weeks, he received an email newsletter from one of the holiday providers with a very attractive offer. Mark clicked on the tracked banner in the email, which took him straight to the offer. Two days later, after consulting with his partner, they made a decision to book. Mark, having previously bookmarked the page, entered it through the bookmark tab and finalized the purchase.
Now the question is: to which channel or medium do we attribute this purchase? When we look into Google Analytics’ standard conversion report, the purchase will be attributed to the email since Mark clicked on the tracked URL and by default, Google Analytics attributes the conversion to the last non-direct click.
One could argue that the purchase wouldn’t have been possible had Mark not found the website on Google Search in the first place. One could also argue that without receiving the email, Mark wouldn’t have known about the special offer and the purchase wouldn’t have been possible.
There are a couple of attribution models available in Google Analytics under the Model Comparison Tool.
- Last Interaction – attributes 100% credit to the last channel. In the above example, Mark’s purchase would have been attributed to direct since Mark entered the website through a bookmark tab on the last visit which ended up with a purchase.
- Last Non-Direct click – attributes 100% credit to the last non-direct In the above example, Mark’s conversion is attributed to the Email, since that was the last non-direct touch point.
- Last Google Ads click – attributes 100% credit for the last Google Ads click. If the holiday provider was running a Paid Search campaign, which would’ve brought Mark to the site and 2 weeks later, he finalized the purchase through email link; Google Ads would get 100% credit for that sale.
- First Interaction – 100% of credit is attributed to the first touch point. In the above example, Mark’s purchase would have been attributed to Google Search (Organic).
- Linear – Each of the touch points along the user’s journey will get equal percentage of credit. In the above example, Organic, Email and Direct would have received 33.3% of credit each.
- Time Decay – The touch point closest to the conversion gets most credit. In the above example, Direct and Email would have received the most credit as opposed to Organic Search, which would have received the least.
- Position Based – 40% of the credit is attributed to the first and last touch points while the remaining 20% is evenly distributed among the middle touch points. In the above example, Organic and Direct would have received 40% of the credit for the purchase, while Email would have been awarded with 20%.
Apart from the above mentioned, a custom attribution model could be setup based on a specific model.
The caveat to the above attribution models is that all of the users’ interactions have to be performed on the same device for the report to be accurate. Why is that? Because Google Analytics sees each unique device (unique set of cookies) as a user.
Going back to the original example – if Mark had searched for the vacation deals on his phone and then finalized the booking on his computer, Google Analytics would see it as two unique users – one Organic (Mobile) who visited the site and another Desktop which Google Analytics would attribute to Direct, since Mark entered the website directly from his bookmark tab. There will be no way for Google Analytics to know that those 2 unique users (or devices) are actually the same physical person.
In a world where mobile usage is becoming more and more common, the above presents a big challenge for analytics. This is particularly apparent with big-ticket purchases, which rarely are finalized on mobiles.
Is there any way a Marketer can attribute actions performed on different devices to a single user? There are.
There are two main approaches for cross-device data-driven attribution, Deterministic and Probabilistic.
Deterministic approach relies on using customer identifiable information. In Google Analytics, users can be identified across multiple devices by implementing a “User ID” feature in your Google Analytics account and website. This obviously requires integration with your database and it goes without saying that the site has to have an account registration option to begin with. Once users login and stay logged in on their devices, it is possible to access cross-device reports allowing visibility in the full user-journey across multiple devices.
Another example would be Facebook with its Pixel. Once implemented on the website it is possible for marketers to pinpoint ad interactions and conversions to the same user, given that they are logged in to Facebook on those devices.
On the other hand Probabilistic relies on algorithmically matching numerous data-points such as geo-location, operating system, resolution and more to find a match between devices. Obviously this method will not be as accurate as the Deterministic approach but some companies claim to have a 70% accuracy rate. Good example of this approach is Cross Device tracking in Google Ads, which relies on combining aggregated and anonymized data which is then used to estimate the number of cross device conversions.
In a world where cross device user journeys will become even more common, accurately attributing a purchase/conversion to relevant channels across multiple devices will always present challenges for marketers. Knowing the limitations of Google Analytics, how and where the data is derived, will prove essential to make sense of it all.
Bookmark Content is a full service global marketing agency and can help you take your digital content strategy, including attribution to the next level.