The role of an online marketer is not just to launch campaigns, it’s also to find out which channels are working the best and which ones aren’t so as to optimize budgets and drive more conversions.
And while the concept of attribution is quite simple – identify which events and touch points contributed most to a conversion throughout a customer’s journey and then assign each event and touch point a certain percentage of credit for that conversion – achieving accurate attribution is still an uphill battle.
One of the most common attribution models is the cross-channel model, which aims to identify which touch points a consumer had across different channels before they converted (e.g. purchased a product or downloaded an ebook). This type of attribution is relatively easy to achieve, provided you implement the right techniques.
However, attribution starts to get challenging when consumers start interacting with a brand across different channels and on different devices.
This is known as the cross-device customer journey.
The Cross-Device Customer Journey
Once upon a time, the online customer journey was undertaken on one device – predominantly desktops / laptops.
Nowadays, consumers use a range of Internet-enabled devices to do a wide range of things – from browsing products on laptops, to searching for flights on tablets, to reading emails on smartphones.
This new generation of multi-device users has given rise to the cross-device customer journey.
According to a study by Google, a majority of online consumers who use multiple devices start their purchase on a smartphone and then continue on a PC or tablet.
Another study conducted by GFK, found the following:
More than 60% of online adults in the U.S and U.K use at least two devices every day, while a quarter (25%) of online Americans and a fifth (20%) of online Britons use three devices.
What’s So Hard About Cross-Device Attribution?
Attribution has never been easy and the increase in multi-device usage just makes things harder for marketers.
The main reason why cross-device attribution is hard to achieve is because there is no accurate way to identify the same user across different devices.
The reason why other types of attribution (e.g. cross-channel and inter-channel) are able to provide reasonably accurate results is because they mainly rely on cookies stored on a consumer’s device (e.g. laptop) to identify and track them – therefore allowing marketers to see each customer’s touch points in their customer journey.
However, cookies are not interchangeable between devices – the cookies on your laptop can’t be transferred to your smartphone or tablet and vice versa – and each device has it’s own way of being identified. For example, desktop users can be identified by cookies and device fingerprints, whereas smartphone users can be identified by their device’s unique ID.
So if marketers can’t rely on cookies as a way to track and identify users across devices, what hope do they have for achieving cross-device attribution?
As it stands, there are 2 main ways to identify one user across different devices – deterministic and probabilistic matching.
What Is Deterministic Matching?
Deterministic matching involves identifying the same user across different devices by connecting the same unique identifiers together.
The most common unique identifier is an email address as it is often used to not only send and receive emails, but to create accounts and log in to different websites and apps.
Take Google Apps for example.
A user could log in to their Gmail account on their smartphone, desktop, and tablet.
This allows Google to deterministically identify the same user across 3 different devices because they use the same email address to access Google Apps on these different devices.
This method of identifying users across different devices is quite accurate (about 80% – 90% accuracy), but it’s mainly reserved for big players – such as Google, Facebook, Amazon, and the like – as they are really the only companies that have a large number of users that actively use their services across different devices.
However, more and more publishers and companies are starting to either encourage (by giving more value/access) or force (by limiting some features/functionality) online consumers to create accounts and sign in to their site and app on different devices. This method does open up the playing field a bit, but it is still limited to large sites (e.g. news sites).
What Is Probabilistic Matching?
Unlike deterministic matching that uses a small number of anonymous unique identifiers to identify the same user across different devices, probabilistic matching uses a range of different data sets and algorithms to make probable connections.
The types of data used in probabilistic matching may include the following:
- IP addresses
- Device IDs
- Browser type
- Interests and web history
- Location
- Language settings
Probabilistic matching also uses deterministic data sets to help train the algorithms (by way of machine learning) to identify the same user across different devices based on their behavior.
What’s Next For Cross-Device Attribution?
In a multi-device, multi-channel world of billions of online and offline consumers, perfecting cross-device attribution will never be achieved – it can only be improved.
FTC in the US and the European Union’s Article 29 Data Protection Working Party (Art. 29 WP) are 2 prime examples of this.
Companies using and improving deterministic and probabilistic matching will have to constantly review privacy regulations and laws to ensure they know which pieces of data they can collect without consent, with consent, or cannot be collected at all.
There are a number of marketing software vendors and analytics companies working hard to improve cross-device identification and attribution, but the key to improving it lies in data, in particular, data management platforms (DMP).
A DMP allows marketers to import and combine all their offline and online first-, second-, and third-party data together to get deep insights into their customers’ behavior, but more importantly, identify users across different devices to improve cross-device attribution and reporting.
