Data Profile for This Category
The core participants in information stream advertising creative testing are information stream ad optimizers, creative teams, and growth teams. Daily produced and used documents include creative scripts, material test sheets, delivery data reports, and iteration checklists. Data sources cover ad platform backends, internal material libraries, user interaction data, and conversion data. The update frequency is daily or weekly. Core data fields include material creative type, delivery time slot, click-through rate, conversion cost, and user interaction feedback tags. Testing work for this category typically revolves around combinations of images, videos, and copy, requiring frequent comparison of performance data across multiple sets of materials.
Constraints Imposed on Workflow Orchestration
The multi-source data characteristics of this category require workflow orchestration to configure multiple independent nodes to pull ad platform delivery data, material library material information, user interaction and conversion data separately, to avoid single-node pull timeouts or data confusion. The daily or weekly update frequency requires the workflow to support both scheduled triggering and manual triggering modes, adapting to the different rhythms of daily iteration and weekly review. The rich data fields and dynamic creative combination requirements require using variables to bind data fields in the workflow, enabling parameterized calls. At the same time, the significant impact of test results requires the workflow to configure a failure alert node to trigger notifications when data is abnormal, ensuring review efficiency and compliance.
Configuration Guidelines
This configuration is based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Workflow trigger timing | Scheduled daily/weekly trigger, or manual trigger | Adapts to the daily/weekly update frequency and flexible testing needs of this category |
| Tool call order | Execute in the order specified by nodeId | Ensures the correct logic chain for data pulling, comparison, and review |
| Variable reference rule | Use variables to bind data source fields | Dynamically adapts to parameter requirements for different test combinations |
| Content truncation threshold | 2000 characters | Filters irrelevant redundant information from generated content |
| Failure alert timeout | 600 seconds | Prevents single-node execution from taking too long and affecting overall workflow efficiency |
| Knowledge base retrieval scope | Only call material libraries and delivery reports related to creative testing | Narrows the retrieval scope to improve review accuracy |
Three Common Misconfigurations
- Phenomenon: Multiple tools configured in the workflow do not execute in the expected order, and some tools are not triggered. Cause: The execution dependency corresponding to
nodeIdis not correctly configured, or the execution order parameter is not declared invariables. - Phenomenon: The AI-generated creative test report includes fixed prompt text at the end, which is not within the preset knowledge scope. Cause: The content truncation rule is not set, or the truncation threshold is set too large, causing the system prompt to be retained as well.
- Phenomenon: After deploying with docker-compose, workflow configurations and knowledge base information are lost, but the interface can be called normally. Cause: A persistent storage volume is not bound in the docker-compose configuration, so data is not retained after the container restarts.
How to Verify Proper Configuration
- Trigger the test workflow, check the execution logs of each tool node, and confirm that the calls are completed in the preset order.
- Check the generated creative test report, and confirm that no irrelevant fixed prompt text is included.
- Verify that the fields bound by
variablesare correctly replaced with the actual content of the corresponding data source, for example, whether the delivery time slot field displays the actual delivery date. - Restart the service, check whether the workflow configurations and knowledge base information are retained, and confirm that there is no loss.
The values provided are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-10-19.