Hey folks, just spent the morning firefighting and I'm seeing some major gaps in our device graph attribution after the latest iOS update. Our usual stitching logic, which leans on a mix of first-party cookies and IDFA with fallback parameters, is showing a ~40% drop in linked sessions for iOS traffic. 😬
Here's a snapshot from our dashboard showing the sudden decoupling. The dip starts exactly correlating with the update rollout.
```sql
-- Simplified query showing linked session rate by OS
SELECT
os_name,
DATE(timestamp) as day,
COUNT(DISTINCT CASE WHEN device_graph_id IS NOT NULL THEN session_id END) * 100.0 / COUNT(DISTINCT session_id) AS linkage_rate
FROM events
WHERE timestamp > NOW() - INTERVAL '7 days'
GROUP BY 1, 2
ORDER BY 2 DESC;
```
We're using a probabilistic model to bridge web and app, but it feels like the ground truth data just vanished for a chunk of users. I'm curious:
* Is anyone else seeing this with their MMP (Mobile Measurement Partner) or home-built attribution stack?
* For those using tools like AppsFlyer, Adjust, or Branch – are you getting alerts on this?
* What's your backup strategy for cross-device tracking now? We're looking at strengthening our first-party data layer and maybe increasing reliance on deterministic web-to-web signals.
Sharing what you're seeing could help us all triangulate the issue faster. I'll update this thread with any config workarounds we find on our end.
Dashboards or it didn't happen.