What the Data for This Category Looks Like
Data for advertising and marketing intelligent due diligence reports comes from files exported from mainstream advertising management systems, third-party ad monitoring reports, and internal brand marketing campaign ledgers. Two update schedules apply: daily campaign data updates on a calendar day basis, and full due diligence data for a single campaign is archived within 7 business days after the campaign ends. The document structure of a single due diligence report includes the campaign identification module, placement channel module, material module, and conversion link module. Fields include campaign ID, channel name, material format, impression count, click count, conversion order count, per-customer consumption amount, and placement time period. Corresponding units are counts, counts, units, yuan, hours, and others.
Constraints for Multi-turn Dialogue and Prompt Engineering
The multi-source, cross-system origin of advertising and marketing due diligence data requires multi-turn dialogue to first confirm the data collection channel and time range. This prevents mixing placement data from different cycles. The complex multi-field structure requires prompt engineering to explicitly specify the campaign identification, channel, and conversion fields to extract. This prevents returning irrelevant, non-due-diligence related content. Association verification rules for different fields—such as impression count being greater than or equal to click count—require multi-turn dialogue to guide users through verifying data rationality step by step. This avoids generating report content with logical contradictions. The archive cycle characteristic of single campaigns requires prompt engineering to support users specifying campaign cycle parameters. This adapts to due diligence query needs across different stages.
Configuration Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
maxContext | 8000–12000 characters | Advertising and marketing due diligence reports contain multi-dimensional associated fields. Multi-turn dialogue must retain multi-round query conditions and data ranges to avoid context overflow |
UPLOAD_FILE_MAX_SIZE | 500 MB | Structured data files for single advertising and marketing due diligence reports typically do not exceed this size, adapting to batch upload requirements |
PARSE_FILE_TIMEOUT_SECONDS | 300 seconds | Parsing structured due diligence files with multiple fields requires extended processing time, preventing premature termination of the parsing process |
maxTurns | 10 turns | Queries for advertising and marketing due diligence typically require 3-5 rounds to refine requirements. Exceeding 10 turns leads to redundant context and reduced response efficiency |
similarity threshold | 0.75 | Precise matching of due diligence dimensions in user queries is required to avoid recalling irrelevant placement history data |
segment length | 1000 characters | Field descriptions in advertising and marketing due diligence reports are clear. Segment length adapts to the information density of field combinations, improving recall accuracy |
The parameter values provided on this page are common starting points for configuration. Actual values are affected by material format, data volume, and business rules. Specific issues require case-by-case analysis. It is recommended to test on your own samples before finalizing values.
Three Common Configuration Errors
- Symptom: After uploading an advertising and marketing due diligence report via the dialogue interface, no parsing progress prompt is displayed, and no error logs appear in the backend. Cause: The uploaded file size exceeds the threshold configured by the
UPLOAD_FILE_MAX_SIZEparameter, and the system skips the parsing process silently. - Symptom: The preset default query question does not load automatically after the dialogue interface starts, and the initial question must be entered manually. Cause: The
default_user_promptparameter is not bound in the dialogue initialization configuration, and the default initial question content is not set correctly. - Symptom: When calling the dialogue interface to generate a due diligence report, the current generation process cannot be actively terminated, and returned results continue to accumulate. Cause: The
stop_sequenceparameter is not enabled, or the correct termination trigger condition is not configured, so stop requests cannot be received.
How to Verify Proper Configuration
- Upload a test file that conforms to the advertising and marketing due diligence report format, check whether the interface displays parsing progress, and whether corresponding parsing logs are generated in the backend.
- Start the dialogue interface, confirm whether the initially loaded question content matches the preset
default_user_promptparameter. - Initiate multi-round queries, call the dialogue stop interface, confirm that the current generation process is terminated, and no additional incomplete results are returned.
- Copy the due diligence report content generated by the dialogue to an external tool, verify whether the format meets expectations, and adjust the format rules in the system prompt.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.