Data Freshness is at the Core of Cross-Device Accuracy

The following is a guest contributed post by Keith Petri, Screen6, Chief Strategy Officer

At Screen6, we have long believed that the solutions in place for cross-device tracking require a fresh set of eyes. It is important that the entire industry that has become reliant on cross-device identification, from buyers to sellers to data processors, understand exactly how any cross-device vendor’s graph is built. This will assure that they can have total confidence in their claims regarding the strength of connections across consumers and devices.

We also believe that marketers need to keep asking questions about the effectiveness of the solutions in market and the quality of the data. Question your current cross-device vendor and question new potential partners, but do not limit yourself to predetermined questions with predefined acceptable answers. In short, asking the right questions will result in better cross-device solutions.

Match rates measure the overall effectiveness of the ID synchronization process – determining that cookie A within a pool belongs to the same person as cookie B from another pool. High match rates between cookies suggest that a particular graph will be more effective and provide access to a larger online audience than a graph with low match rates. Today, match rates between asynchronous cookie pools are what the industry relies on to determine the effectiveness of most cross-device ID vendors. Not all vendors work this way, but some providers still depend on syncing disparate cookie pools.

Marketers use these match rates as a method of validation and comparison among these types of master device graph providers. To truly understand match rates, marketers need to be asking about the types of data that go into a graph, how the match rate claim is calculated and who is vetting the claim. Most importantly we need to consider how fresh the data is and how frequently a graph is updated and sent to clients.

Currently, most cross-device providers build and send their ID graphs to clients once every seven to ten days. This timeline and overall approach to providing actionable data is a core issue that impacts the match rates in cross-device identification, if they define match rates correctly. How big this problem of data freshness and cookie decay rates poses is something we have unique insight into.

We studied trillions of server-to-server events to conduct this analysis. Every day we process each of our clients’ datasets for not only the past 24 hours, but looking back over a variable amount of time. As such, we can identify the cookie (ID) depreciation rate. While some vendors have cookies with significantly higher longevity in lifespan (i.e. >14 days), our analysis shows that the average half-life of cookies across the billions we see daily and trillions we see monthly is 6 to 8 days.

With the average half-life sitting around 1-week, cookies depreciate at the same rate which the average cross-device vendor refreshes its graph associations. If you see a new user on Monday, and a cross-device vendor who only refreshes its graph weekly, on Sundays, returns an updated mapping file the next day (Monday, one-week later) – then what percentage of the graph is actually viable to be leveraged for targeting?

Marketers today need to think both in real-time and in 24-hour increments, reconsidering how often they review and manage data and their campaigns. The chain of old and bad data will have a negative impact on campaign results. We encourage clients to look into their ID graphs and examine the percentage of the graph which matches not just to inventory which they have seen, but inventory which they see after the delivery date of the graph. This is an analysis that most platforms find surprising.

Advertising technology prides itself on advancing the field; everything from profiling and segment creation to ad serving and tracking. Our job is to create new solutions that meet the new demands of the market and the job of marketers is to ask the questions that cut through false claims about bad data, old data and match rates and dig into the core attributes of a cross-device graph.