How does cross-device behavior break attribution?
Still have a question, spotted an error, or have a better explanation or a source we should cite?
Email attribution is built on the assumption that the device you click on is the device you convert on. In reality, that's often not true, and the gaps are significant.
Here's a common scenario: someone opens your email on their phone during their commute and clicks a product link. They like what they see but don't buy on the phone. Later that evening they search for the product on their laptop, land on your site directly, and purchase. From a purely cookie-based attribution standpoint, email gets no credit. The laptop visit came direct. The phone click is forgotten.
Why? Cookies don't transfer between devices. The session on the phone and the session on the laptop are treated as different users unless something connects them. Most analytics platforms can't make that connection without a logged-in user identifier.
Some platforms get around this with probabilistic cross-device matching (inferring that two devices are the same person based on behavior patterns) or deterministic matching (matching via a logged-in account across devices). Logged-in e-commerce experiences and apps handle this better. Anonymous website visitors are much harder.
The practical implication: your email conversion rate and ROI are almost certainly understated because of cross-device breakage. You're not getting credit for conversions that email influenced but that completed on a different device. This is a good reason to set a slightly longer attribution window than you think you need, even if it picks up some noise at the edges, and to rely on cohort-level analysis rather than last-click attribution where possible.
Contributors
Who worked on this answer
Every name links to their profile. Every company links to their site. Real people, real accountability.
