Data Analysis

Anomaly Diagnosis

AE User ID Binding Diagnosis

Diagnose user ID fragmentation caused by multi-endpoint reporting in AE, automatically detect Account ID and Distinct ID missing rates, locate the root cause of fragmentation, and output a code-level fix.

IndustriesAll industries

Agent · Skills Live

skill · loadedindustry+100

Run this Skill on my data

Agent orchestration

multi-agent

Analysis readyevidence attached

AE User ID Mapping Anomaly Investigation is a data-quality capability in ThinkingAI's industry Skill library. It helps data engineers diagnose user ID fragmentation caused by multi-endpoint reporting (client plus server) in AE, automatically detect the missing rates of Account ID and Distinct ID, distinguish normal business behavior from abnormal fragmentation, locate the root cause, and output a code-level fix. This is ThinkingAI's decade of industry experience, packaged into ready-to-use Skills.

Live walkthrough

AgentSave as AgentExport session

te-user-id-binding-diagnosis

Why can't the same player's logins and payments be tied to one user? The designer confronted me with the retention report, saying the numbers are noticeably lower than what it feels like. We report from both client and server, so help me check whether the ID stitching chain is broken somewhere.

Pull the coverage of #distinct_id / #account_id / #user_id across the last 7 days of events

Client events miss #account_id 25% of the time, server events miss #distinct_id 30% of the time

Distinguish normal business from abnormal fragmentation: exclude multi-character login scenarios

Multi-character is only 3.2%; after removing it the miss rate barely changes, so this is abnormal fragmentation

Split the miss distribution by reporting source (client SDK / server API)

SQL samples and compares the login / payment event ID mapping chain for the same device

In broken samples, login and payment fall under two different #user_ids

Pinpoint the break point, generate a code-level fix

Investigation done. It's not AE's ID-merging rules that went wrong, it's that each of the two reporting ends is missing a leg: client events miss #account_id, server events miss #distinct_id, and the same player got split into two #user_ids, with logins on A and payments on B, so retention naturally comes out low.

1. ID stitching chain, checked link by link

Chain link	Reporting source	Check item	Result
Device initialization	Client	#distinct_id generation	Normal · 100% coverage
Login callback	Client	#account_id binding	Break point 1 · 25% missing
Server reporting	Server	#distinct_id pass-through	Break point 2 · 30% missing
ID merging	AE system	#user_id generation	Rule normal, but input incomplete

← Scroll to see more

2. Attribution chain

After a successful client login, the binding call isn't made immediately → the login event carries only #distinct_id → the server's payment reporting in turn carries only #account_id → AE never gets a single event containing both IDs to merge on → the same player generates two #user_ids → the login chain and payment chain each get attributed to a separate person. At the current miss rate, retention calculation is off by about 3–5 percentage points, and behavior-sequence completeness is only 70%, and the designer's gut feeling is right.

The fix touches just two spots of code: (1) on the client, bind `#account_id` immediately in the login callback, don't wait for the next event; (2) on the server, pass through `#distinct_id` from the logged-in state when reporting. Both are changes at the SDK call layer and don't involve any AE configuration.

Recommendation

Based on the fix benchmark for comparable dual-end reporting projects: the #account_id miss rate can drop to 0.5% and the #distinct_id miss rate to 1%, with behavior-sequence completeness recovering from 70% to 98%. A note on scope: this Skill locates the break point and gives reporting-code fix recommendations, but won't change the live SDK for you; after the fix ships, I'd suggest re-running this diagnosis once to validate, and whether the already-fragmented historical data needs recomputing should be assessed separately.

The diagnosis report and the two fix code snippets are synced to the data team's group, with validation SQL included.

On your data

That was a simulated run

Leave your work email and we will run a live walkthrough on your real business data.

The problem

User identity in AE relies on three fields: #distinct_id, #account_id, and #user_id. In multi-endpoint reporting, about 50% of dual-endpoint projects have ID-missing problems, usually with a missing rate of 15% to 40%. Client events have #distinct_id but lack #account_id, and server events have #account_id but lack #distinct_id, splitting one user into multiple records and directly affecting retention, channel attribution, and behavior sequences.

What it does

Distinguish normal from abnormal: first determine whether it is normal business behavior (such as multi-character login) or abnormal fragmentation (such as server reporting missing an ID), avoiding misdiagnosis

Automatic SQL detection: automatically run SQL to query the Account ID and Distinct ID missing rates and quantify the severity of fragmentation

Code-level fix: not only diagnose the problem but also provide SDK reporting-code changes, resolving ID fragmentation at the root

When to use it

Investigating user ID fragmentation in a multi-endpoint reporting scenario

Diagnosing when the Account ID missing rate is abnormally high in AE

Investigating channel-attribution skew caused by missing Distinct ID

Verifying ID reporting completeness after a new SDK version launches

Investigating ID mapping when a user's behavior sequence shows a break

In the field

Case

A game project · client and server reporting ID break

The same user's login and payment events could not be linked to one #user_id. The Skill detected a 25% #account_id missing rate on client events and a 30% #distinct_id missing rate on server events. The fix was to bind #account_id immediately in the client login callback and pass through #distinct_id on server reporting. Afterward the Account ID missing rate fell to 0.5%, the Distinct ID missing rate to 1%, and behavior-sequence completeness rose from 70% to 98%.

FAQ

Does multi-character login count as ID fragmentation?

No. Creating multiple characters on one device is normal business behavior. The Skill distinguishes normal from abnormal first, and only fragmentation caused by missing reported IDs enters the investigation.

Does the fix change SDK code or AE configuration?

Usually it changes SDK reporting code, ensuring the client binds account_id after login and the server passes through distinct_id on reporting.

How much does ID fragmentation affect the data?

It affects retention analysis, payment attribution, and channel analysis. 10% ID fragmentation can skew the retention-rate calculation by 3 to 5 percentage points.