We've been running Tealium for event collection for about 18 months. The visitor stitching is now creating significant profile duplication in our downstream CDP, inflating our identified user count by an estimated 15-20%. This breaks our SLA reporting for audience activation.
Our core issue: anonymous visitor IDs from `utag_main` are being merged with authenticated user IDs inconsistently. We see the same user with multiple merged profiles.
Current setup:
* `utag_main` cookie for anonymous ID.
* `user_id` set via `utag.link` on login.
* Tealium's Visitor API is enabled for ID merging.
The duplication metrics are clear. Over a 7-day period:
* Total tracked events: 4.2M
* Events attributed to "merged" profiles: ~700k
* Distinct merged profile IDs generated: ~85k (should be closer to ~50k based on our auth logs)
Has anyone deconstructed this flow and forced a stricter merge? We're looking at the `tealium_visitor_id` logic, but the documentation is vague on edge cases.
What we've tried:
* Ensuring the `user_id` is stable and sent before any other events post-login.
* Verified there are no race conditions in the `utag.link` call.
Is the merge truly deterministic, or are we at the mercy of the order of events hitting the Data Hub? We need concrete rules, not marketing speak.
Metrics don't lie.