What this category's data looks like
Enterprise sales product training materials consist of three categories: product manuals, sales response scripts, and channel training schedules. Sources are the product department, sales operations, and channel management, with update cadence following version releases and quarterly schedule adjustments. When materials enter the knowledge base retrieval and recall function, product manuals serve as the basis for parameters and functional boundaries, sales response scripts serve as the basis for scenario responses, and channel training schedules serve as the basis for timing and personnel arrangements. Fields across the three material types include product name, feature item, script number, schedule date, channel name, and signed receipt. During retrieval, material fragments are used as candidate sources, relevance is determined by similarity scores, and the final cited content is passed to subsequent nodes or manually used to generate a handover-ready deployment status and operation records.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate fragments and similarity scores (after pattern, merge, rerank, filter, and limit trimming). The handover-ready deployment status and operation records are then completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Product manual | Whether fields, dates, and signatures are consistent with the sales response scripts and channel training schedules | In the search test, switch searchMode to semantic, full-text, or hybrid, compare candidate fragments and similarity scores, and verify product name, quantity, date, and signed receipt |
| Sales response script | Whether script numbers and product names align with the product manuals and channel training schedules | Test once each with synonym-rewritten questions, check whether candidates include script numbers and product names, and record the number of candidates and scores |
| Channel training schedule | Whether schedule dates, channel names, and signed receipts are complete | Verify schedule dates and channel names, check whether signatures or receipts are missing, and compare the ranking of the top few items before and after enabling rerank |
| Deployment status and operation records | Whether version, connection, and task records are consistent | Check version, connection, and task records item by item according to the handover form, and record the candidate order before and after the usingReRank switch and the cited text tokens after limit trimming |
Three common errors
- Symptom: The same product name is written differently in the product manual and the sales response script, and dates do not match. Cause: The two materials were not updated in sync, lacking unified field verification.
- Symptom: The channel training schedule has dates but no signed receipts, and the handover form is missing items. Cause: The signature process was not closed-loop, and receipts were not archived.
- Symptom: There are many candidates but the top-ranked ones are irrelevant, and the answer lacks key paragraphs. Cause: The retrieval mode does not match the material format, and no rerank secondary sorting was performed.
How to verify the setup
Acceptance for this function: Test once each with synonym-rewritten questions, compare candidate fragments and similarity scores, then toggle rerank to observe the ranking of the top few items, and record the number of candidates and scores.
Business test: Check item by item once according to the handover form, see which record is missing, and verify the fields, dates, and signatures of the product manuals, sales response scripts, and channel training schedules.
- In the knowledge base search test, switch
searchModeto semantic, full-text, or hybrid, and compare candidate results and similarity scores. - Compare the order of the top few items before and after enabling rerank, check the impact of
usingReRankon ranking, and record the cited text tokens afterlimittrimming.