New digital tools that connect customer journeys across multiple devices are the hottest new toys in ad tech. Done well, they can uncover true campaign value that has been hidden by single-device tunnel vision.
Yet cross-device solution providers have struggled to find accessible language to explain what they do, instead focusing on marketing jargon.
A good cross-device partner will demonstrate how complex most customer journeys are, will be honest and transparent about the limitations of their technology, and work with you to prove any performance gains.
Before you explore the space, familiarise yourself with the top ten terms you’ll need to know to meaningfully discuss cross device with a technology partner.
Deterministic versus probabilistic cross-device solutions
A supplier can match two devices by tracking the same visitor across the web (deterministic matching), or create a predictive algorithm based on their extensive online traffic data (probabilistic).
Training data set
For probabilistic solutions, this is the data your cross-device partner uses to build their model. The device-to-user matching predictions are generalised based on the data used in the calibration stage. Cross-device partners usually have billions of data points to make predictions across demographics and markets.
A device graph is the pairing of cookies sent back after the matching algorithm does its magic. It’s basically cookie pairings that belong to the same user. For probabilistic device graphs there may also be a measure of how confident the supplier is that the algorithm accurately predicted the pairing.
Precision is the percentage of devices correctly identified by the supplier’s device graph out of the total predictions made. Precision is usually calculated on a subset of the training data of the supplier or on your own CRM and cookie data. Deterministic models have near perfect precision as they do not have any statistical error.
Whereas precision is about correct numbers divided by total predictions made, recall is the correct number as a percentage of all actual visitors and their multiple devices (and cookie ids). This is a measure of relevance. Deterministic data has much lower recall since it is relevant only to the exact visitor’s data.
Device partners will want to sync cookies by placing their pixel into one of your tracking pixels (also know as ‘piggy-backing’). They can then associate their cookie id with your cookie ids and produce a device graph.
This is the percentage of your cookies passed to the provider. It’s usually between 80%-90% after an initial cookie sync, and you should work with a provider to make this as high as possible.
The percentage of your cookies the device mapping has associated with a device. You should see a 40-60% rate for probabilistic models and 10-40% for deterministic. Careful: this is not a reliable measure of accuracy. Instead, focus on high recall and precision.
A walled garden is a provider of cross-device solutions solely within its all-in-one content, advertising and analytics platform. Facebook is one. If you rely on Facebook’s cross-device targeting in their advertising products you’ll get a cross-device solution that is packaged and relevant to that platform’s audience but not transparent (FB won’t report on match rates, precision and recall for example).
Dissynergy means the fall in quality of, or a discontinuation in, services after a merger or acquisition. Most of the cross-device graphs are
provided by smaller tech companies (or start-ups). Any future acquiring company might change priority of the cross-device product stack to the disadvantage of existing customers.
If you understand the above terms, you’ll be set up to create meaningful relationships with cross-device solution providers, beyond simply seeking out one of the hottest, and usually quite expensive trends.