In our last article, we talked about web-tracking and the different ways we collect visitor data to understand how visitors interact and engage with a website. The realities around web-tracking can leave gaps in the user journey and can lead to inaccurate reporting. For example, cookies may tell us that there are three different people visiting a website. However, ‘people-based’ or ‘identity-based’ measurement can identify that it is in fact one person visiting using three different devices. Given this journey/cookie fragmentation, how can marketing professionals link the number of visits with each individual to ensure they are delivering better brand experiences to their customers?
The first step to help solve the problem is to piece data sources together. This process is called ‘stitching’.
Stitching combines the multiple visits into one profile where we know it is the same person. These ‘visits’ are actually really just one person on different devices rather three different people. It’s an important differentiation. Why should marketing professionals pay for three people when really, it’s only one? In addition, stitching allows brands to gain a better, more accurate understanding of customer journeys so that customers get more relevant content and more personalized experiences.
One of the most common stitching methods is ‘probabilistic stitching’. This relies on pulling data from syndicated pools. These pools store cookies or ID’s that have been bought from companies. Algorithms then make calculations based on probabilities and match up different user ID’s to what they think is likely to be the same person.
While this method is often more accessible to smaller companies because it requires less time and infrastructure to set up, there are some obstacles and limitations. Firstly, it is difficult to verify the accuracy of the data or audit it because you don’t have access to the raw data itself. Secondly, in Europe, it is now almost always in breach of the new GDPR regulations.
The other method is ‘deterministic stitching’. This uses information that customers have agreed to share with your brand. It may simply be a unique identifier (e.g. internal user ID, e-mail) that is matched against the raw data (cookies or another identifier). It combines the information into an individual profile based on data-driven insights to be sure they are the same person. There are no probabilities or assumptions involved. There are some limitations with deterministic models however. It’s often harder for smaller companies with a more basic infrastructure to keep people logged in, which can be a challenge with a deterministic strategy.
At Fospha, we apply an advanced data-driven deterministic method. We work with companies using their first party data to stitch and overlay additional data sources such as order references, email addresses and offline user data. Our systems are fully compliant with GDPR legislation and we are able to measure and attribute customer journeys, creating highly accurate profiles and audience clusters which in turn allow for more personalized content customization. Interest level data means brands have more insight into customers so that they can share more relevant content and offers to improve brand experience and increase customer engagement.
Our measurement methodology allows us to differentiate between two different individuals who may log in on the same device. We are experts in helping brands stitch previously disconnected data together to create hyper accurate profiles and audience clusters. We do all this from a purely data-driven and consent led perspective, removing probabilistic assumptions. Fospha’s advantage also lies in our ability to be flexible and agile in client environments and is highly customisable.